1145232040 Q * ntrs__ Ping timeout: 480 seconds 1145232240 J * ntrs ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1145232730 Q * ntrs Ping timeout: 480 seconds 1145233121 J * ntrs ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1145234479 J * manolito ~almacen@host-200-58-77-58.supernet.com.bo 1145234580 M * Bertl welcome manolito! 1145234611 M * Bertl as probably everybody figured by now, the EHegg net was not the best :/ 1145234613 M * manolito thanks 1145234723 M * Bertl will back finally back in an hour I guess ... 1145234729 N * Bertl Bertl_oO 1145236888 P * manolito 1145238154 N * Bertl_oO Bertl 1145238159 M * Bertl back now ... 1145238170 M * Bertl hey perfect estimation :) 1145238190 M * daniel_hozac haha, indeed. 1145238254 M * Bertl daniel_hozac: how are you? everything fine? 1145238272 M * daniel_hozac yep, how was easterhegg? how are you? 1145238313 M * Bertl was fun, not very organized but what can you expect from a ccc event 1145238343 M * daniel_hozac ccc? 1145238402 M * Bertl chaos computer club 1145238424 M * daniel_hozac lol 1145238559 M * Bertl btw, the loopback ip stuff I tested looks very promising :) 1145238594 M * daniel_hozac hmm? 1145238607 M * Bertl ah, didn't I mention (yet)? 1145238623 M * daniel_hozac i must've missed it if you did. 1145238635 M * Bertl maybe I didn't .. was late yesterday 1145238646 M * Bertl I basically made 127.0.0.1 work inside a guest 1145238656 M * Bertl without rewrite :) 1145238713 M * daniel_hozac so each guest has their own 127.0.0.1? 1145238721 M * Bertl yup 1145238739 M * daniel_hozac nice. 1145238747 M * Bertl still needs a few changes, and of course testing, testing testing ... 1145238757 M * daniel_hozac i guess that will please a lot of folks. 1145238787 M * Bertl yeah, it came up more than once recently 1145238808 M * Bertl so I thought about using some of the ngnet techniques for that 1145238847 M * Bertl and while it wont give you tcpdump-ability 1145238875 M * Bertl it can ease the local ip thing and of course ifconfig now with the recent changes 1145238893 M * daniel_hozac yeah. 1145238912 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net 1145238946 M * Bertl wb matta! 1145238974 M * Bertl any folks around who are eager to test linux-vsever network stuff? 1145239085 M * daniel_hozac nebuchadnezzar? 1145239435 M * Bertl probably not right now .. seems almost everyone is asleep :) 1145239462 M * Bertl even kenelnewbies only sees joins and leaves 1145239484 M * Hollow hm 1145239490 M * Hollow morning :) 1145239509 M * Bertl good morning Hollow! 1145239520 M * Hollow good night would fit better :) 1145239593 M * Bertl k, good night Hollow! :) 1145239615 M * Hollow no, i'm not leaving.. just came home from partying.. 1145239622 M * Hollow and still too awake 1145239635 M * Bertl hmm, welcome back from party then *G* 1145239643 M * Hollow :) 1145239697 Q * matta Ping timeout: 480 seconds 1145239781 M * Hollow Bertl: what needs to be tested? 1145239947 M * Skram hi all 1145239977 M * daniel_hozac hello 1145239977 M * Bertl Hollow: well, I have a highly experimental and somewhat incomplete patch to allow for 127.0.0.1 inside the guests (in a secure way) 1145240019 M * Bertl and I would like to know how it behaves with real guests 1145240034 M * Hollow real as in production or as in a real distro 1145240044 M * Bertl the latter 1145240059 M * Hollow ok, good, i wouldn't sacrifice the former :D 1145240060 M * Bertl only tested it with qemu and netwotk contexts yet 1145240124 M * Bertl it should not explode (at least not immediately) but it will definitely contain some bugs (missing stuff) 1145240192 M * Hollow ok.. just have to install some sources first 1145240210 M * Hollow ah.. i'm lucky, rc16 is already there 1145240235 M * Hollow patch? 1145240241 M * Bertl sec 1145240683 M * Bertl (longer sec) 1145240691 M * Hollow :) 1145240770 M * Bertl against 2.6.16-vs2.1.1-rc16 I presume? 1145240776 M * Hollow yup 1145240982 M * Bertl hum, no idea but my diff takes ages ... investigating now 1145241068 M * Bertl but it's still running and doing stuff ... 1145241078 M * daniel_hozac built tree? :) 1145241097 M * daniel_hozac (i've done that too many times...) 1145241111 M * Bertl it's quite strange, it is doing the right thing 1145241131 M * Bertl but it seems that the kernel tree was purged out of the inode cache 1145241139 M * daniel_hozac ah. 1145241303 M * Bertl okay, it's a little mixed with recent 'other' changes, but I guess that doesn't really matter 1145241359 M * Bertl http://vserver.13thfloor.at/Experimental/delta-lo0.02.diff 1145241410 M * Hollow ok, compiling now 1145241454 M * Bertl btw, did you see the 'improved' PLM tests? 1145241463 M * Hollow no? 1145241553 M * Bertl https://plm.osdl.org/plm-cgi/plm?module=patch_info&patch_id=5091 1145241684 M * Hollow hm, have to use a vanilla 2.6.16.. only have the gentoo one 1145241699 M * Hollow +5 min 1145241712 M * Bertl np, let me clean up the patch in the meantime 1145241719 M * Hollow hm.. what is this PLM? 1145241735 M * Bertl kernel cross compile tests, basically 1145241739 M * Hollow ah 1145241829 Q * daniel_hozac Quit: brb 1145241849 J * daniel_hozac ~daniel@c-2d1472d5.010-230-73746f22.cust.bredbandsbolaget.se 1145241913 M * Hollow Bertl: any special kconfig i could need? 1145241938 M * Bertl we are not OVZ :) no, but you should enable networking :) 1145241960 M * Hollow :P 1145241965 M * Bertl debugging might be helpful too 1145241993 Q * softi42 Ping timeout: 480 seconds 1145242085 M * Bertl okay, updated the patch in place 1145242111 M * Bertl will now only contain lo stuff and avoid flodding you with debug messages 1145242620 J * softi42 ~softi@p549D5731.dip.t-dialin.net 1145242729 M * Bertl wb softi42! 1145243339 Q * shedi Quit: Leaving 1145243839 M * Hollow Bertl: ok.. for the guest i need no extra config? 1145243855 M * Bertl no, just make sure that it has only non local IPs 1145243864 M * Bertl i.e. not 127.0.0.1 assigned 1145243867 M * Hollow ok 1145243872 M * Bertl and I only tested with one IP actually 1145243886 M * Bertl but it should work for more than one too 1145244892 Q * Hollow Quit: Konversation terminated! 1145244913 M * Bertl hmm, not sure that is a good sign :/ 1145245193 J * Hollow ~hollow@home.xnull.de 1145245246 M * Hollow works. 1145245248 M * Hollow :) 1145245262 M * Bertl wow, sounds good :) 1145245270 M * Bertl let's hear about the details! 1145245289 M * Hollow i need to find them out first, but lo is already there 1145245314 M * Bertl ah, well, that is the easy part :) 1145245324 M * Bertl interesting are the following tests: 1145245339 M * Bertl - contact a guest service, on the guest via 127.0.0.1 1145245361 M * Bertl - bind one service in several guests to 0.0.0.0 1145245375 M * Hollow ok 1145245377 M * Bertl - check that they can be reached via 127.0.0.1 and you actually get the right one 1145245394 M * Bertl - bind a service explicitely to 127.0.0.1 only 1145245408 M * Bertl - bind a service in several guests to 127.0.0.1 1145245434 M * Bertl if possible, do all of this on udp and tcp 1145245447 M * Bertl btw, does ping on 127.0.0.1 work inside? 1145245457 M * Hollow hm.. do you know a small udp daemon to test with? 1145245468 M * Bertl tinydns? 1145245472 M * Bertl just kidding :) 1145245481 M * Bertl nc should do the trick (netcat) 1145245696 M * Hollow 1) bind different daemons to 127.0.0.1 on host and guest works 1145245735 M * Hollow (just installing netcat) 1145245785 M * Hollow wait 1145245796 M * Hollow binding to 0.0.0.0 in guest + 127.0.0.1 on host works 1145245803 M * Hollow but binding both to 127.0.0.1 fails 1145245821 M * Hollow now i get EADDRINUSE on the host 1145245844 M * Bertl ah, okay 1145245861 M * Bertl so you bind to 127.0.0.1 in the guest (first) then the host cannot bind that, right? 1145245887 M * Hollow exactly 1145245901 M * Bertl yeah, I missed that case ... 1145246384 M * Hollow it doesn't work vice versa as well 1145246396 M * Hollow first bind 127.0.0.1 on the host, then guest fails 1145246431 M * Bertl okay, likely the same issue, let's forget about the host for now 1145246456 M * Hollow ok.. 1145246529 M * Hollow contact a guest service, on the guest via 127.0.0.1: works 1145246607 M * Hollow hm, if i bind to 0.0.0.0 in a guest, it only binds to the ip i configured in util-vserver 1145246657 M * Hollow hm no 1145246662 M * Hollow with nc it works 1145246667 M * Hollow maybe i should stick with nc :P 1145246758 M * Hollow hm 1145246761 M * Hollow very strange 1145246780 M * Bertl udp vs tcp maybe? 1145246827 M * Hollow no.. if i specify a port with nc it binds to the configured ip only.. if don't give it a port it uses a random one and binds to 0.0.0.0 1145246916 M * Bertl hmm, but you use it with -l, right? 1145246925 M * Hollow yep 1145246978 M * Hollow localhost ~ # nc -l 0.0.0.0 &>/dev/null & 1145246982 M * Hollow [1] 3900 1145246986 M * Hollow localhost ~ # ss -l 1145246990 M * Hollow Recv-Q Send-Q Local Address:Port Peer Address:Port 1145246994 M * Hollow 0 0 *:38551 *:* 1145246998 M * Hollow localhost ~ # killall nc 1145247002 M * Hollow localhost ~ # nc -l -p 1234 0.0.0.0 &>/dev/null & 1145247007 M * Hollow [2] 3905 1145247010 M * Hollow [1] Exit 1 nc -l 0.0.0.0 >&/dev/null 1145247014 M * Hollow localhost ~ # ss -l 1145247018 M * Hollow Recv-Q Send-Q Local Address:Port Peer Address:Port 1145247020 M * Hollow 0 0 192.168.0.123:1234 *:* 1145247020 M * Hollow ops 1145247020 M * Hollow http://phpfi.com/113157 1145247141 M * Bertl what is ss again? 1145247154 M * Hollow iproute2's equivalent for netstat iirc 1145247175 M * Bertl hmm, I have iproute2 but no ss here ... 1145247193 M * Hollow well, netstat -nl --inet shows the same 1145247218 M * Bertl which means we do not see that there is anything bound, right? 1145247238 M * Hollow we see that it bound to 192.168.0.123 when it should to 0.0.0.0 1145247261 M * Bertl ah, have to scroll to the right *blush* 1145247301 M * Hollow well, my terminal windows is so big *g* 1145247318 M * Bertl wasn't that a virus? :) 1145247325 M * Hollow :P 1145247338 M * Hollow s/windows/window/ (for the protocol) 1145247361 M * Bertl could you do strace -fF's of both of them? 1145247403 M * Bertl and what if you do 'nc -l -p 1234' 1145247411 M * Bertl instead of specifying the 0.0.0.0 1145247501 M * Hollow the same as with 0.0.0.0 1145247555 M * daniel_hozac net/ipv4/af_inet.c:inet_bind, shouldn't the if (s_addr == 0) block vanish too? 1145247620 M * Hollow Bertl: http://home.xnull.de/misc/nc1.log 1145247622 M * Hollow and nc2.log 1145247633 M * Hollow nc1 = without -p 1145247636 M * Hollow nc2 = with -p 1145247821 M * Bertl daniel_hozac: hmm, yes and no, we need some 'special' check there, but the current one is probably wrong 1145247905 M * Bertl correction, it is definitely wrong :) 1145247911 M * daniel_hozac :) 1145247940 M * Bertl fascinating, it is full of bugs :) 1145247946 M * Bertl + s_addr2 = 0xffffffffl; 1145247968 M * Bertl shouldn't that be the broadcast address of the context? 1145247978 M * daniel_hozac it will be. 1145247999 M * daniel_hozac hmm, yeah, that assignment needs to moved into the if (nxi) block. 1145248012 M * daniel_hozac (the s_addr2 = v4_bcast;) 1145248096 M * Bertl I think I will rip out the entire network checks and replace them, keeping only the address/socket/interface comparisons 1145248146 M * Bertl but not right now :) 1145248175 M * Hollow *nod* i'm leaving to bed 1145248206 M * Bertl thanks a lot Hollow! 1145248213 M * daniel_hozac me too. good night all. 1145248221 M * Hollow you're welcome, hope to see it in rc17 :) 1145248230 M * Bertl second that ... cya tomorrow! 1145248240 N * Bertl Bertl_zZ 1145248252 M * Skram tootles. 1145248257 M * Skram i wish i could go to sleep 1145248263 M * Skram god. I am on the phone with a above.net tech, he couldnt read the server labels, then got all pissed at me because he has to configure eth0, pfft. 1145250039 Q * kilian Remote host closed the connection 1145250136 J * kilian kk@projects.verfaction.de 1145250163 Q * nebuchadnezzar Ping timeout: 480 seconds 1145252619 J * kir_home ~kir@swsoft-mipt-nat.sw.ru 1145254844 M * tokkee Good morning ;-) 1145254873 M * tokkee Which vserver version would you recommend for a production environment? 1145255869 J * Viper0482 ~Viper0482@p549750BF.dip.t-dialin.net 1145256629 Q * kilian Read error: Connection reset by peer 1145258268 J * nebuchadnezzar ~nebu@zion.asgardr.info 1145258412 J * kilian kk@projects.verfaction.de 1145259485 J * SAHUNG SAHUNG@sw73-173-242.adsl.seed.net.tw 1145259598 J * dapp1111 ~test123_1@203-73-171-6.adsl.dynamic.seed.net.tw 1145261729 J * bonbons ~bonbons@83.222.37.206 1145263276 Q * cryo Ping timeout: 480 seconds 1145263334 M * h01ger moin moin 1145263802 M * nebuchadnezzar lo 1145264153 J * Oli ~skycode@ip-213-49-144-133.dsl.scarlet.be 1145264380 J * dapp1 ~test123_1@203-73-171-6.adsl.dynamic.seed.net.tw 1145264450 Q * dapp1111 Read error: Connection reset by peer 1145264813 J * dapp2 ~test123_1@203-73-174-144.adsl.dynamic.seed.net.tw 1145265001 Q * nebuchadnezzar Quit: ERC Version 5.1.2 $Revision: 1.796.2.4 $ (IRC client for Emacs) 1145265093 J * nebuchadnezzar ~nebu@zion.asgardr.info 1145265128 Q * dapp1 Ping timeout: 480 seconds 1145266105 J * cryo ~say@psoft.user.matrix.farlep.net 1145266720 Q * Hollow Quit: Konversation terminated! 1145266759 J * Hollow ~hollow@home.xnull.de 1145266821 J * |coocoon| ~coocoon@p54A068B6.dip.t-dialin.net 1145266851 M * |coocoon| hello 1145266990 Q * dapp2 Ping timeout: 480 seconds 1145268150 M * phreak`` bonbons: got a sec ? 1145268178 M * bonbons yep, can interrupt IPv6 hacking for a sec ;) 1145268187 M * phreak`` heh, great :) 1145268198 M * phreak`` bonbons: about http://bugs.gentoo.org/show_bug.cgi?id=99778 :) 1145268203 M * phreak`` (its pretty old :P) 1145268243 M * phreak`` guess it still doesn't compile with hardened-gcc, right ? 1145268291 M * bonbons was a long time ago, think I also hit something with hardened some time ago, let me read the bug first 1145268352 M * bonbons but since then I didn't retry with hardened on 1145268508 M * bonbons would be worth a try I guess 1145268555 M * phreak`` well it seems to be related (if not the same) to http://bugs.gentoo.org/show_bug.cgi?id=114796 (the -nopie thing) 1145268602 M * phreak`` because if you compile diet with -nopie; the utils also compile/work fine 1145268622 M * bonbons isn't nopie a hardened-related flag? (introduced in gcc-3.3*, 3.4* by hardened) 1145268631 M * phreak`` bonbons: yeah .. 1145268663 M * bonbons remember vaguely something like that 1145268669 M * phreak`` (pie as in position indepenent executeables) 1145268847 M * Hollow phreak``: lol, you know what? 1145268866 M * phreak`` Hollow: hm ? :) 1145268888 M * Hollow debian packages a /usr/lib/diet/lib/libc.so 1145268910 M * Hollow instead of the libc.a 1145268937 M * Hollow would be interesting to see their build script 1145268948 M * phreak`` so I really should take a closer look at the debian-dietlibc :) 1145268991 M * Hollow i didn't say that 1145268992 M * Hollow :P 1145269003 M * Hollow make dyn 1145269004 M * phreak`` but I did ;) 1145269017 M * Hollow well, maybe it has advantages... but who knows 1145269166 Q * |coocoon| Quit: KVIrc 3.2.0 'Realia' 1145269321 M * bonbons I will let you try out by now, I should really progress a bit with IPv6!! (CoW helps a lot on it - ultra-quick diffs, newly discovered make O=builddir for kernel helps too) 1145269599 M * nebuchadnezzar trying to start a vserver I get chroot-shopen("/etc/mtab"): Permission denied 1145269599 Q * cryo Read error: Connection reset by peer 1145269631 J * cryo ~say@psoft.user.matrix.farlep.net 1145271483 M * nebuchadnezzar arf, when I tried to exclude /usr/src from hashification I put this to /etc/vserver/.default/apps/vunify 1145271508 M * nebuchadnezzar I think it the reason for /etc/* to have UI attr 1145271697 M * nebuchadnezzar does /etc/vservers//apps/vunify/exclude add exclude to the default list of does it completely replace it ? 1145271985 M * nebuchadnezzar arf, it completely replace it 1145272078 M * nebuchadnezzar should be intersting to have /etc/vservers/.defaults/apps/vunify/exclude replace the default list, but /etc/vservers//apps/vunify/exclude only add or remove some 1145272450 Q * SAHUNG Ping timeout: 480 seconds 1145272680 Q * mire Ping timeout: 480 seconds 1145273538 J * |coocoon| ~coocoon@p54A06A23.dip.t-dialin.net 1145273863 Q * |coocoon| Quit: KVIrc 3.2.0 'Realia' 1145274027 J * mire ~mire@193-166-222-85.COOL.ADSL.VLine.Verat.NET 1145274033 J * |coocoon| ~coocoon@p54A06A23.dip.t-dialin.net 1145274045 Q * Oli Ping timeout: 480 seconds 1145274825 Q * phedny Ping timeout: 480 seconds 1145275337 J * phedny ~mark@volcano.p-bierman.nl 1145275910 J * kir ~kir@swsoft-mipt-nat.sw.ru 1145276263 J * brc bruce@20151219045.user.veloxzone.com.br 1145276727 Q * Viper0482 Ping timeout: 480 seconds 1145276856 Q * cehteh Remote host closed the connection 1145277309 J * Viper0482 ~Viper0482@p549768AF.dip.t-dialin.net 1145278006 J * lilalinux_ ~plasma@dslb-084-058-216-169.pools.arcor-ip.net 1145278420 Q * lilalinux__ Ping timeout: 480 seconds 1145279229 Q * |coocoon| Quit: KVIrc 3.2.0 'Realia' 1145279452 M * bonbons Bertl_zZ: How does it work inside kernel to call functions that may be located in modules? (like call it only if module loaded and prevent linker errors) 1145279672 M * nebuchadnezzar shuuuttt, he's sleeping 1145279680 A * nebuchadnezzar make no noise 1145279682 M * nebuchadnezzar :-) 1145279799 Q * cryo Ping timeout: 480 seconds 1145280075 J * cryo ~say@psoft.user.matrix.farlep.net 1145280131 J * shedi ~siggi@cpe-24-165-167-167.midsouth.res.rr.com 1145280134 J * Ljungbacka ~ljungback@nl117-191-93.student.uu.se 1145280818 M * nebuchadnezzar context are coded with 4bytes ? 1145280992 M * bonbons nebuchadnezzar: context ID you mean? 1145280999 M * nebuchadnezzar yes 1145281002 M * nebuchadnezzar they are 2bytes 1145281020 M * nebuchadnezzar I test to launch a vserver with contextID 65540 > it fails 1145281181 M * bonbons it's xid_t == unsigned int 1145281191 M * nebuchadnezzar ha 1145281213 M * nebuchadnezzar why can not run a vserver with 70000 as context ? 1145281224 M * bonbons anyhow, if 65539 succeeds it can't be 16 bit! 1145281298 M * nebuchadnezzar bonbons: unsigned int is 4bytes on x86 ? 1145281317 M * bonbons hardcoded limit, and don't forget that somewhere around 40000 (if I remember well) there's the limit to dynamic IDs 1145281400 M * nebuchadnezzar hmm hmm 1145281423 M * bonbons yep sizeof(unsigned int) == 4 on x86 1145281497 M * Hollow nebuchadnezzar: xid_t is uint, but there is an arbitrary limit MAX_S_CONTEXT in the kernel 1145281523 M * nebuchadnezzar bonbons: does it have something to do you TagID ? 1145281540 A * nebuchadnezzar is lost 1145281552 A * nebuchadnezzar try to understant the limit :-) 1145281557 M * bonbons in include/linux/vserver/context.h: #define MAX_S_CONTEXT 65535 /* Arbitrary limit */ 1145281561 M * Hollow as i said, it's arbitrary 1145281588 M * nebuchadnezzar ok 1145281590 M * nebuchadnezzar so 1145281592 M * nebuchadnezzar ok 1145281748 M * bonbons and for the dynamic contexts: #define MIN_D_CONTEXT 49152 1145282213 M * nebuchadnezzar ok, now 1145282242 M * nebuchadnezzar the vhasify exclude, I see in the doc that it's rsync-like 1145282264 M * tokkee Can anybody tell me what the struct nameidata * argument in vfs_mkdir is for? 1145282388 M * nebuchadnezzar but it do not seems to work like rsync 1145282588 A * nebuchadnezzar slaps himself 1145283424 M * bonbons tokkee, isn't that BME-related? 1145283468 M * bonbons like info about mountpoint and such? (having a look at fs/namei.c) 1145283545 M * tokkee bonbons: What is BME? 1145283598 M * bonbons patch to allow RO bind-mounts 1145283636 M * bonbons that is, each fs command needs to know on what mountpoit it's operating (to check if the fs isn't read-only or has other limitations) 1145283688 M * tokkee Sounds like it is BME-related ;-) 1145283718 M * tokkee However changing the signature of such important function might break a lot of other things imho... 1145283729 M * tokkee s/function/functions/ 1145283759 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net 1145283777 M * tokkee bonbons: Do you know what might be a good default value to pass as struct nameidata argument? What happens if I simply pass NULL? 1145283779 M * bonbons ask Bertl for the details, he's BME-patch maintainer for a long time, if not even original author (I'm not sure) 1145283831 M * tokkee Well, I wrote an email to the list - I guess he will respond ;-) 1145283832 M * bonbons I think Berl said it doesn't hurt to pass in null, as that's what is used in mainline at lower levels 1145284039 M * tokkee Hmmm... there are a couple of functions (e.g. xid_permissions) that accesses the nameidata argument without checking if it's NULL - if I pass NULL to such a function, I'd get a NULL pointer exception (don't know what will happen in the kernel)! 1145284074 M * bonbons possibly a panic... let me check 1145284115 M * bonbons fs/nfsd/vfs.c:1163: err = vfs_mkdir(dirp, dchild, iap->ia_mode, NULL); 1145284175 M * tokkee There are a couple of places in the kernel sources, where NULL is passed to some of those functions... 1145284217 Q * cryo Ping timeout: 480 seconds 1145284300 M * bonbons looks like the may_* functions are there to catch it, but I'm digging 1145284412 M * tokkee vfs_mkdir() passes the nameidata argument to may_create() (w/o checking for NULL) which passed it on to permission() (no checking again) which again w/o checking passes it on to xid_permission()... 1145284434 M * tokkee => crash ;-) 1145284442 M * tokkee Or did I miss an important line? 1145284480 Q * matta Ping timeout: 480 seconds 1145284698 M * bonbons in permission: (nd && MNT_IS_RDONLY(nd->mnt))), nd IS checked for null 1145284789 M * tokkee bonbons: Have a look at line 273... 1145284791 M * bonbons the vxwprintk in dx_permission looks dangerous though 1145284821 M * tokkee bonbons: imho we could reach that point without checking for NULL. 1145284932 M * bonbons yep, we should tell Bertl_zZ once he's back. Let's see if there are more such dangerous lines 1145284993 M * tokkee Btw. is there a way to tell (by checking a define or something like that) if the kernel sources contain the VServer patch? 1145285082 M * bonbons should be possible, CONFIG_VSERVER_... or negate version of it 1145285261 M * tokkee Something like that would be necessary to check which version of the function to call in a driver/addon (like e.g. unionfs). 1145285407 M * bonbons you could also check for existance of IS_IMMUTABLE() and such macros in include/fs.h (just need to find a BME-related one so you get it right also on older vserver versions) 1145285461 M * tokkee Well... vserver 2.0.1 did not change the signature of all those functions... 1145285581 M * bonbons look at http://vserver.13thfloor.at/Experimental/BME/ for BME patches. I've seen in there mount.h, MNT_IS_RDONLY define 1145285599 M * Hollow tokkee: well you could just chek if the vserver syscall exists ;) 1145285616 M * bonbons tokke: Bertl merged BME into vserver only recently, there was a long thread about it on ML some time ago 1145285675 M * bonbons Hollow: but that does not always detect BME, or did I miss something? 1145285698 M * Hollow no, it does not, but he asked how to find out if the kernel has the vserver patch 1145285701 J * matta ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1145285749 M * tokkee Hollow: Well the correct question would be how to detect BME ;-) 1145285750 M * bonbons but it's the change to vfs_* function signature which is in cause <= BME in vserver 1145285795 M * Hollow hm, you could check VCIVersion.. but that will only tell if you have a version with bme included, and probably not even that 1145285897 M * tokkee What does VCIVersion stand for? 1145285917 M * Hollow vserver api version 1145285972 M * tokkee Hmmm... sounds like a good solution to me ;-) 1145285988 M * Hollow well, not sure that the version changed with the inclusion of bme 1145286041 M * tokkee :-( 1145286092 M * bonbons checking for definition of MNT_IS_RDONLY in mount.h sounds efficient. (there are a few other new defines by BME patch) 1145286102 M * Hollow mhm 1145286116 M * Hollow depends when you need to check 1145286122 M * Hollow i assume for kernel modules 1145286124 M * Hollow ? 1145286132 M * tokkee ack 1145286163 M * Hollow yeah, then the check for define should be ok 1145286231 M * tokkee Alright... thx. 1145286256 M * tokkee Now, I just need to know if I may savely pass NULL to the vfs_* functions ;-) 1145286334 J * Oli ~skycode@212.224.238.239 1145286369 M * tokkee I'll be gone now.... cya ;-) 1145286375 M * Hollow cya 1145286396 M * bonbons cya 1145286466 M * h01ger hmmmmmmmmm... in my firewall on the server i can see some packets coming from 127.0.0.1 but i know those are originating from a vserver inside - how come ? 1145286533 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net 1145288291 M * blizz hm, is resident set size memory the "surely in use" memory? ;-) 1145288978 P * pollux 1145289602 M * gdm resident set size is the virtual memory pages currently in ram, from what i understand 1145289865 Q * kir_home Ping timeout: 480 seconds 1145291088 Q * bonbons Ping timeout: 480 seconds 1145291142 M * daniel_hozac h01ger: yeah, the source address isn't rewritten. 1145291250 M * daniel_hozac h01ger: i've got a patch for it, but i haven't had a chance to test it yet. (and i guess it might not be relevant if the guest isolated loopback works out) 1145291272 A * h01ger nods 1145291312 J * bonbons ~bonbons@83.222.37.206 1145293520 N * Bertl_zZ Bertl 1145293525 M * Bertl morning folks! 1145293547 M * bonbons Hi Bertl! 1145293584 M * phreak`` morning Bertl :) 1145293707 M * gdm hey Bertl :) 1145293826 M * Bertl hey bonbons, phreak``, gdm! and others! :) 1145293868 M * Bertl bonbons: what kind of module functions are you talking about? 1145293915 M * bonbons it's "addrconf_lock()" and "in6_dev_finish_destroy()" the linker complains of (for IPv6) 1145293985 M * bonbons methods invoked by checks used to determine net devices that should be visible to guest (I took IPv6 equivalents of your IPv4 implementation) 1145294076 M * Bertl why does it complain? 1145294109 M * bonbons e.g. undefined reference to `addrconf_lock' 1145294139 M * bonbons that is when linking vmlinux 1145294155 M * Bertl ah, well, then you missed to include some header 1145294162 M * Bertl which probably defines that one inline 1145294178 M * Bertl you should get a warning from the compiler for that one too 1145294188 M * Bertl (for every occurance, that is) 1145294303 M * bonbons there is no warning, just the linker error at the end 1145294327 M * Bertl that sounds unusual ... 1145294368 M * Bertl but it's possible, if you 'forgot' to links some file containing that function 1145294375 M * Bertl *link 1145294445 M * daniel_hozac will that even work if IPv6 is compiled as a module? 1145294475 M * Bertl yes, if the appropriate parts are define external and all dependacies are met 1145294483 M * bonbons I'm compiling IPv6 as a module, did not yet try with compiled-in IPv6 1145294535 M * Bertl daniel_hozac, bonbons, @all: do you know good tools to measure net socket performance? 1145294623 M * bonbons Bertl: but to work at runtime I need to check if the module is loaded, otherwise the call will go to na-mans-land. I don't want to pull in IPv6 myself in this case 1145294693 M * nebuchadnezzar I make some test of vserver and 2.6.7-rc1 1145294711 M * bonbons grep says: ./net/ipv6/ipv6_syms.c:32:EXPORT_SYMBOL(in6_dev_finish_destroy); and /include/net/addrconf.h:147:extern void in6_dev_finish_destroy(struct inet6_dev *idev); 1145294767 M * daniel_hozac http://kernel-perf.sourceforge.net/about_tests.php ? 1145294812 M * nebuchadnezzar when booting, mounting xfs over the LVM tell me that barrier is not handle by underling device 1145294921 M * Bertl bonbons: you don't want to check for a loaded module and call into nowhere at runtime 1145294947 M * Bertl check out how the vanilla kernel does that with ipv6 loaded or unloaded 1145294995 J * cryo ~say@psoft.user.matrix.farlep.net 1145294996 M * bonbons ok, will have to find a code sample in vanilla. :) 1145295036 M * daniel_hozac does the kernel call any IPv6 functions if IPv6 isn't loaded? doesn't IPv6 register itself somehow? 1145295060 M * Bertl daniel_hozac: tx, netperf looks like what we want 1145295114 M * bonbons daniel_hozac: you are possibly right, but some modular code certainly does what we want 1145295187 M * bonbons Bertl: did you already see tokkee's fs/BME-concerns in backlog? 1145295285 Q * Oli Quit: Oli 1145295294 M * Bertl I did answer to hos email on the ML 1145295298 M * Bertl *his 1145295335 J * MrX ~urk@219.95.239.112 1145295347 M * Bertl welcome MrX! 1145295358 M * MrX Hi 1145295490 M * bonbons Bertl: ok, but there seem to be dangerous parts in code: some of the wrapped debugging printk's dereference nb without checking if it's NULL! 1145295517 M * Bertl ah, okay, missed that, reading up now 1145295635 M * bonbons one found to the end of dx_permission() in namei.c 1145295689 M * bonbons back later, eating something 1145295699 M * Bertl enjoy your meal! 1145295731 M * bonbons thanks! 1145296340 J * flosch ~flosch@e178245204.adsl.alicedsl.de 1145296346 M * Bertl welcome flosch! 1145296394 M * flosch hi 1145296578 J * Oli ~skycode@212.224.238.239 1145296583 M * Bertl wb Oli! 1145296644 M * Oli hi all 1145296764 M * Bertl tokkee: nice catch, what about this one: http://vserver.13thfloor.at/Experimental/delta-ndpath-fix01.diff 1145296801 M * Bertl tokkee: btw, do you have a test case which actually triggers it? 1145297512 M * Bertl Oli: you had a question yesterday, no? was it answered? (I don't remember it right now) 1145297516 J * mkhl ~mkhl@200-148-41-116.dsl.telesp.net.br 1145297523 M * Bertl wb mkhl! 1145297855 J * eyck eyck@ghost.anime.pl 1145297863 M * Bertl wb eyck! 1145298153 M * bonbons Bertl: I don't think there was any testcase, was while reading to find out some info about the new nb argument and checking if null is possible 1145298817 J * kevinp ~kevinp@ny.webpipe.net 1145298826 M * Bertl welcome kevinp! 1145298838 M * kevinp thanks 1145298941 M * matta Bertl: i swear that's an onjoin script you're using :) 1145298983 M * Bertl matta: well, you are wrong ... 1145299019 M * matta you're just that good! :) 1145299083 Q * Ljungbacka Quit: Peace and Protection 4.22.2 1145299186 M * Bertl matta: thing is, I see no good way to convince folks (like you) that it is no script/trigger, and actually I don't want to spend time on that either ... 1145299235 M * matta lol was just joking Bertl :) 1145299286 M * Bertl yeah, but for you the joke is quite old ... 1145299355 M * kevinp I created a post-start script for vserver disk space limits that works great when I run it manually as root, but when it is run automatically by executing `vserver test.com start` (also as root), I get "permission denied" errors. 1145299385 M * kevinp does the vserver start script run as a different user? 1145299404 M * Bertl hmm, it should not, but some scripts run _inside_ the guests (context) 1145299413 M * daniel_hozac kevinp: you know that 0.30.210 supports disk limits in the configuration, right? 1145299426 M * Bertl daniel_hozac: really? 1145299428 M * kevinp wow, no hadn't heard that 1145299446 M * daniel_hozac it's one of the two major changes in it ;) 1145299450 M * kevinp I just upgraded to .210 to see if it was fixed, but am still running it the old way 1145299453 M * Bertl daniel_hozac: how does it do the summing? 1145299458 M * daniel_hozac vdu. 1145299483 M * Bertl so it's a proper and full featured solution. great! 1145299522 M * Bertl fascinating, totally missed that one ... 1145299627 M * bonbons Bertl: I think I will continue patching for IPv6 putting module IPv6 aside (to be fixed later by moving the problematic function part into ipv6 module and calling it directly as a single symbol function) [but will have to understand the mutex implications] 1145299646 M * kevinp daniel_hozac, looks like it was your patch, just found it in the changelog, nice job 1145299647 M * Bertl bonbons: I'd suggest doing so for now 1145299755 M * bonbons If you have a good source for safe call to (maybe not-loaded) module (that does not load the module on-demand!) it could be of use! But it's not blocking for now 1145299771 M * Bertl daniel_hozac: I think the entire inet_bind() check can go, except for the in nx info test, no? 1145299815 M * Bertl (in the lo and no single ip bind case) 1145299856 M * daniel_hozac yeah, i guess so. 1145299969 Q * michal` Ping timeout: 480 seconds 1145300106 M * Bertl daniel_hozac: funny thing, devinet_notiproot() seems to be already dead code :) 1145300298 M * daniel_hozac hehe, indeed. 1145300355 Q * phedny Ping timeout: 480 seconds 1145300358 M * Bertl I wondered about the name, because I remembered some ifa checks 1145300608 J * the_hydra ~a_mulyadi@202.59.168.5 1145300668 M * Bertl welcome the_hydra! 1145300688 M * the_hydra hello bert! 1145300692 M * the_hydra happy easter! 1145300705 M * the_hydra still zzzz now? ;) 1145300723 M * Bertl and a happy one to you too .. no, not sleeping anymore :) 1145300730 M * the_hydra hehehe 1145300787 M * the_hydra getting vacation and went somewhere during the past holidays? :) 1145300823 M * Bertl no, just work and talks ... 1145300830 M * the_hydra hmm 1145300853 J * phedny ~mark@volcano.p-bierman.nl 1145300870 M * the_hydra btw, bert, does Linux kernel account the page cache hit ratio? 1145300940 M * Bertl you mean some kind of statistics about page misses vs hits? 1145300948 M * the_hydra yeap 1145300964 M * Bertl well, the hits are hard to figure, the misses are accounted 1145300986 M * the_hydra hm..maybe I missed that...where I can look it up? 1145300999 M * Bertl # time true 1145300999 M * Bertl 0.00user 0.00system 0:00.01elapsed 0%CPU (0avgtext+0avgdata 0maxresident)k 1145300999 M * Bertl 0inputs+0outputs (90major+17minor)pagefaults 0swaps 1145301004 Q * Viper0482 Remote host closed the connection 1145301010 M * Bertl for example shows major and minor pagefaults 1145301031 M * softi42 hi - i finaly managed to get a vserver with network up and running. I want ro tun apache inside. 1145301041 M * softi42 but abache complains: 1145301042 M * the_hydra Bertl: oh.."time".. 1145301087 M * softi42 Could not determine the server's fully qualified domain name. Does anyone know how apache determins this name? 1145301130 M * Bertl probably by either reading stuff from /etc/hosts or name services or looking at the apache config file 1145301186 M * Bertl softi42: a 'host' or nodename of test.example.org is fully qualified, test OTOH, is not 1145301236 M * softi42 i'll try /etc/hostname - then /etc/hosts ... 1145301560 M * the_hydra gtg bert 1145301563 M * the_hydra cu later 1145301568 M * Bertl the_hydra: cya! 1145301570 Q * the_hydra Quit: 1145302367 Q * Wonka Ping timeout: 480 seconds 1145302562 M * softi42 one more problem - after i run vserver test1 stop - my network intreface has no more ip adress (i am using eth0) 1145302587 M * Bertl does your eth0 have one before you start the guest? 1145302600 M * softi42 yes 1145302619 M * Bertl does the guest have the same ip? 1145302619 M * softi42 i get the ip using dhcp 1145302631 M * softi42 yes 1145302661 M * Bertl does your config dir (/etc/vservers//interfaces/0) contain a 'nodev' entry? or a dev entry? 1145302699 M * softi42 /etc/vservers//interface contains a dev entry 1145302766 M * softi42 ok - i fond the documentation 1145302792 M * softi42 i should have a nodev etry if i use the primary interface - right 1145302793 M * Bertl okay, you want to put a nodev entry there and remove the dev entry 1145302806 M * Bertl (when you reuse the host ip) 1145302904 Q * oliwel Quit: Chatzilla 0.9.71 [Firefox 1.5.0.1/2006021318] 1145303229 M * softi42 ok - apache starts and can be accessed ;) 1145303253 M * Bertl excellent! 1145304891 N * Dr4g_ Dr4g 1145305604 J * doener ~doener@i5387E20A.versanet.de 1145305651 M * Bertl welcome doener! 1145305667 M * doener evening folks, Bertl! 1145305695 M * kevinp daniel_hozac: so now that I have created the files in dlimits, to implement them, do I need to reboot the vserver or is there another way? 1145305712 M * daniel_hozac yeah. 1145305786 M * daniel_hozac you could of course set it manually. 1145305815 M * kevinp right, just checking if there was a "reload" option or something I didn't know about 1145305840 M * kevinp so what's this new cache feature for? 1145305843 Q * flosch Ping timeout: 480 seconds 1145305902 M * kevinp just a way to increase speed when querying for info about the vservers? 1145305905 M * daniel_hozac so you won't have to wait for 5 minutes every time you start the vserver. 1145306033 M * kevinp so on restart I get two processes that don't die, irqbalance and minilogd, are these even needed to start? 1145306066 M * Bertl irqbalance is now userspace? 1145306079 M * Bertl kevinp: no, IMHO not at all 1145306094 M * kevinp I wouldn't think so, I think I just left them in on accident before 1145306320 J * derjohn ~derjohn@80.69.37.19 1145306333 M * Bertl evening derjohn! 1145306377 M * derjohn juhu! My xorg is back! Nice to be here again1 1145306410 M * derjohn Bertl, what about the Logo? 1145306430 M * Bertl well, I assume you heaven#t got any reply yet? 1145306439 M * derjohn nope. 1145306464 M * Bertl we should wait until tomorrow evening though, as folks probably get back from easter holidays 1145306466 J * dearaujo ~dan@pixpat.austin.ibm.com 1145306474 M * Bertl welcome dearaujo! 1145306477 M * derjohn Or should I simply enlarge the pixel-based one? 1145306498 M * Bertl derjohn: nah, that look really ugly when printed 1145306552 M * derjohn depends, but anti-aliases edges -especially after scaling- indeed look f....d. 1145306651 M * bonbons Ok, going to sleep, will continue IPv6 hacking tomorrow evening, did some nice progress today :) 1145306698 M * derjohn bonbons, v6 without NGnet? 1145306711 M * bonbons yep, like current v4 1145306728 M * derjohn bonbons, then: sleep weel and get fit soon ;) thx ! 1145306768 M * bonbons thanks, will need some good inspiration! 1145306809 M * derjohn bonbons, inspiration or motivation? 1145306842 M * bonbons inspiration to solve IPv6 as module issue 1145306894 M * bonbons need some functions from IPv6 module in kernel itself, that's not a simple call 1145306915 M * derjohn bonbons, well, why simply tell menuconfig that v6 as module and vserver enabled exclude each other? 1145306927 M * derjohn s/why/why not/ 1145307007 M * bonbons because linker is the first one to complain. Then if I call some address in no-mans-land, that will just be panic. So need to lookup module, get symbol if module loaded and call the symbol at that time 1145307031 M * daniel_hozac where are you calling IPv6 functions from? 1145307034 M * bonbons but I currently dont know the details (especially relating to race-conditions with module unloading) 1145307082 M * derjohn bonbons, I would say I sounds liek you need Java Reflections :) Sadly I cannot help you with that particular C issue, I am not a native speaker ;) 1145307085 M * bonbons that's when listing interfaces in nx context, scanns all interfaces and only shows those with an IP (v4 or v6) in the context 1145307108 M * tokkee Bertl: No... don't have any test-cases so far. 1145307115 P * dearaujo 1145307131 M * tokkee Bertl: The patch looks fine - I don't know if there are any more critical sections, though. 1145307132 M * Bertl tokkee: I think it's pretty hard to trigger 1145307152 M * daniel_hozac bonbons: hmm? like addr_in_nx_info? 1145307188 M * bonbons caller of that: dev_in_nx_info() 1145307240 M * daniel_hozac bonbons: can't you just put that in an #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)? much like all of those functions are in an #ifdef CONFIG_INET? 1145307275 M * daniel_hozac eh, nevermind that. 1145307280 M * daniel_hozac i don't know what i was thinking. 1145307310 M * bonbons you were just thinking what I thought a few hours ago :-) 1145307460 M * bonbons let's see what inspirations night brings! 1145307481 M * daniel_hozac yeah, they say sleeping is when you have your brightest moments :) 1145307539 M * Bertl tokkee: if you find other places where this could be an issue, please let me know 1145307592 M * daniel_hozac include/linux/module.h:symbol_get? 1145307620 M * bonbons that's the area I'm considering 1145307657 M * tokkee Bertl: Of course I will ;-) 1145307697 M * kevinp anyone know where to stop minilogd from starting on a redhat based guest? 1145307705 M * bonbons daniel_hozac: need to find out what the word 'mutex' in some comments implies (modules.c), would be bad to get the module removed while calling its functions 1145307709 M * daniel_hozac kevinp: start a syslog. 1145307709 M * tokkee Bertl: Maybe I'm able to construct a case to trigger a panic ;-) 1145307764 J * Wonka produziert@chaos.in-kiel.de 1145307773 M * Bertl tokkee: well, would be nice, but not worth spending more than a few minutes on it 1145307781 M * Bertl wb Wonka! 1145307790 M * Wonka thx 1145307807 M * Wonka i want to work with experts, one day only... 1145307845 M * Wonka chaos.in-kiel.de is a vserver and was down, because s/o created a vserver with the same IP as the vserver host 1145307850 M * Wonka and started it 1145307859 M * Wonka _and_ stopped it 1145307870 M * Wonka which took down the whole eth0 1145307878 M * daniel_hozac hehe. 1145307891 Q * bonbons Quit: of to bed, good coding! 1145307896 M * Wonka any possibility to ad _some_ sort of safeguard there? 1145307909 M * Wonka why is eth0 taken down if there's still secondaries? 1145307909 M * h01ger Wonka, chpasswd ? :) 1145307921 M * Wonka h01ger: no, legitimate root user... 1145307934 M * tokkee Bertl: ack 1145307939 M * Bertl Wonka: because the network stack folks consider that 'proper' behaviour 1145307953 M * Wonka Bertl: they suck. 1145307957 M * Bertl Wonka: but after months of ranting, there is now a config option for that 1145307958 M * tokkee ;-) 1145307965 M * Bertl Wonka: but it is off by default 1145307972 M * tokkee Good night folks ;-) 1145307978 M * Wonka Bertl: what's it named? 1145308001 M * Bertl daniel_hozac: probably remembers :) 1145308003 M * daniel_hozac net.ipv4.all.promote_secondaries, IIRC. 1145308014 M * Bertl lol 1145308019 M * daniel_hozac +conf. 1145308020 Q * derjohn Ping timeout: 480 seconds 1145308095 M * Wonka thx 1145308110 M * kevinp so this is weird, I start syslog, there is no minilogd running, but when I execute a restart it still times out and blames minilogd 1145308174 M * daniel_hozac probably because your syslog shuts down too early. 1145308194 M * daniel_hozac (i.e. something logs after syslog has died) 1145308321 M * kevinp So I moved K88syslog to K99syslog so it is last before the killall and still the same result 1145308483 M * daniel_hozac is the killall doing anything that would cause logging? do your logs show anything after the "exiting on signal 15"? 1145308590 M * ray6 reee from munich... 1145308596 M * daniel_hozac if you strace -fF /etc/rc.d/rc 0 inside the guest, you might get some clues too. should at least tell you which process spawns it. 1145308637 M * kevinp I'll try that next, the next line after the 15 is the syslogd restart line 1145308691 M * brc Bertl: hey, i am waiting to continue de tests 1145308713 M * Bertl excellent! I have a patch for your ready 1145308798 M * kevinp so the strace repeats the following two lines: 1145308806 M * kevinp poll([{fd=3, events=POLLIN|POLLPRI}], 1, 500) = 0 1145308813 M * kevinp stat64("/dev/log", {st_mode=S_IFSOCK|0700, st_size=0, ...}) = 0 1145308830 M * Bertl brc: http://vserver.13thfloor.at/Experimental/delta-cq0.01.diff 1145308838 M * daniel_hozac if you could upload the output somewhere, that would be great. 1145308857 M * Bertl brc: that one is supposed to work as before, i.e. you should not see any difference 1145308876 M * Bertl brc: if you get any kernel panic or unusual behaviour, please notify me immediately 1145308953 M * brc ok 1145309028 M * brc linux-2.6.16.1-vs2.1.1-rc15-delta 1145309033 M * brc on 2.1.1.-rc15 as i was using before 1145309044 M * brc Do i have to change anything on make config ? 1145309048 M * Bertl no 1145309088 M * brc ok 1145309162 M * kevinp daniel_hozac: http://dev.webpipe.net/strace.0.out 1145309235 M * daniel_hozac kevinp: are you sure you use -fF? 1145309249 M * kevinp let me check 1145309259 M * kevinp strace -fF /etc/rc.d/rc 0 > /root/strace.0.out 2>&1 1145309296 M * daniel_hozac well, FYI, strace has an -o option. 1145309327 M * kevinp sorry, I try, but I'm not a programmer by any means, the entire output is in greek :) 1145309338 M * Bertl kevinp: np, take your time 1145309352 M * daniel_hozac i guess initlog is your problem. 1145309427 M * kevinp Bertl: I just hate to waste your time, I know every minute I use on the IRC is one less minute you get to work on evrything else 1145309483 M * Bertl kevinp: don't worry, that's only a very tiny task (of those I do in parallel :) 1145309508 M * kevinp lol, here's the output using -o http://dev.webpipe.net/strace2.0.out 1145309627 J * derjohn ~derjohn@80.69.37.19 1145309720 M * ray6 ree derjohn 1145309752 M * derjohn hm, let see how long radeon/r200/r300 let me stay here ... (since xorg R7) 1145309756 M * derjohn ray6, hey ;) 1145309778 M * ray6 derjohn: use an irc-bouncer in an vserver somewhere :) 1145309833 M * derjohn ray6, well doent help much if cant see whats happening ;) ... but i should care about .. hmm ... what's cool nowadays: psybnc ? 1145309891 M * daniel_hozac kevinp: does running that leave a minilogd process? because i can't see where it would be created... 1145309893 M * ray6 derjohn: oh I have no idea, my irc client runs directly on a stable host :) 1145309913 M * derjohn ray6, then: thanks for the tipp :/ 1145309938 M * kevinp daniel_hozac: yep, it's there alright 1145309939 M * daniel_hozac derjohn: i just switched to irssi inside screen. 1145309947 M * ray6 derjohn: i just meant: when your workstation is too unstable to stay on the channel connect from somewhere else :) 1145309954 M * doener derjohn: hm, xorg R7 is rock stable on my t43 (r300 chipset) 1145309983 M * daniel_hozac kevinp: what does /proc//exec point to? 1145310042 M * daniel_hozac -c 1145310049 M * brc bertl i got an error 1145310057 M * derjohn well, it maybe with xorg original driver which is sloooow on my dualhead (merged framebuffer). so i switchec to r300 from freedesktop.org, then to r200, now I disabled dri (I need the driver only for 2d speed ....) 1145310066 M * derjohn doener, which driver do you run? 1145310068 M * kevinp daniel_hozac: /proc/22926/exe -> /sbin/minilogd 1145310074 M * brc i did a make clean and trying again... 1145310132 M * daniel_hozac kevinp: and it wasn't running before? 1145310263 M * kevinp right, I can reboot and try it again, checking before and after 1145310289 M * daniel_hozac very strange... 1145310347 J * matt1 ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1145310351 M * kevinp "syslogd -m 0" is running before I run it 1145310407 M * kevinp let me post a fresh output, the one I posted was after I had already run it previously 1145310423 M * kevinp (just realized that would make a difference) 1145310433 M * daniel_hozac hehe. 1145310541 M * kevinp ok, third times the charm: http://dev.webpipe.net/strace3.0.out 1145310574 M * kevinp you'll notice that the last two lines keep repeating and still haven't stopped writing 1145310628 M * daniel_hozac that's probably minilogd. 1145310815 Q * matta Ping timeout: 480 seconds 1145310905 M * daniel_hozac so it seems rc is trying to log after syslog has shutdown. 1145311031 M * Bertl okay, here is the next try: http://vserver.13thfloor.at/Experimental/delta-lo0.03.diff 1145311031 M * daniel_hozac does IN_INITLOG=1 /etc/rc.d/rc 0 work? 1145311054 M * Bertl have to leave now, SO just returned ... will be back shortly ... 1145311058 N * Bertl Bertl_oO 1145311351 M * Oli gnite guys 1145311403 M * kevinp that appeared to work 1145311427 Q * Oli Quit: Oli 1145311429 M * kevinp I get this error on all of them though: /etc/init.d/functions: line 373: 21: Bad file descriptor 1145312026 M * daniel_hozac what distro is that? CentOS? 1145312471 J * michal` ~michal@www.rsbac.org 1145312707 Q * mkhl Quit: 1145313293 M * Wonka http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.7 1145313410 M * derjohn Wonka, only one fix for a .7 ? is that so critical? 1145313418 M * Wonka derjohn: looks like. 1145313424 M * Wonka derjohn: also look at .6 and .5 1145313440 M * Wonka derjohn: also only one fix each 1145313461 M * derjohn nox, .6 contains more stuff 1145313516 M * Wonka oh, yes. hadn't really read it, it seems 1145313535 M * derjohn wff is MADV_REMOVE ? 1145313581 M * kevinp daniel_hozac: yeah, centos 1145314572 J * Aiken ~james@tooax6-043.dialup.optusnet.com.au 1145315227 J * lilalinux__ ~plasma@dslb-084-058-195-174.pools.arcor-ip.net 1145315463 J * Omegafire ~Omegafire@CPE0004acd9195c-CM014080213127.cpe.net.cable.rogers.com 1145315494 Q * Omegafire Quit: 1145315655 Q * lilalinux_ Ping timeout: 480 seconds 1145315997 Q * matt1 Ping timeout: 480 seconds 1145316201 N * kevinp kevinp_gone 1145316232 Q * SNy Remote host closed the connection 1145316241 J * SNy 884dfbacff@bmx-chemnitz.de 1145317091 M * waldi wtf? 1145317091 Q * insomniac Read error: Connection reset by peer 1145317255 M * daniel_hozac waldi? 1145317308 M * waldi daniel_hozac: 2.6.16.7 1145317317 M * daniel_hozac ah, yes. 1145317346 P * Dr4g Open Source Development :: http://dynamichell.org 1145317571 J * tkooda ~tkooda@216.70.31.95 1145317578 M * tkooda hello all 1145317584 M * tkooda anyone alive? 1145317616 M * daniel_hozac yep. 1145317638 M * tkooda how does vserver differ from selinux? 1145317668 M * daniel_hozac they are widely separate entities. 1145317691 M * daniel_hozac vserver lets you run multiple isolated servers on one physical host. 1145317764 M * tkooda so, starting a vserver pretends the kernels already running (cause it uses the running parent kernel), chroots (etc), and then runs /sbin/init as if it were a new machine booting? 1145317791 M * daniel_hozac well, how it starts depends on the configuration, but pretty much, yeah. 1145317803 M * tkooda basically just chroot-on-steroids? (oversimplification, but still) 1145317811 M * daniel_hozac right. 1145317838 M * tkooda and selinux is hooks to limit particular actions w/out nessicarily chrooting.. /me thinks 1145317885 M * tkooda I need to run proprietary software with older libs, so I'm thinking vserver'll solve my task 1145317899 M * tkooda ..but still keep the host clean 1145317901 M * daniel_hozac sure. 1145318362 M * derjohn n8 folks, I'll be in S3 mode now ....