1115510548 M * Doener` basically we 'suffer' from the fact that the pid is mapped to the inode#, right? 1115510580 M * Bertl yup, we 'should' have a separate proc for each context 1115510589 M * Bertl (that would be the best solution) 1115510598 M * Doener` but not very efficient ;) 1115510616 M * Bertl hmm .. well depends ... 1115510633 M * Bertl IMHO it would not really add any overhead or require more resources 1115510670 M * Bertl already thought about coding the xid into the inode number for proc entries 1115512665 M * Doener` Bertl: got a short explanation what a negative dentry lookup is? 1115512721 M * Bertl sure .. the dentry lookup is based on a path 1115512741 M * Bertl which is basically considered 'existing' except for the last entry 1115512751 M * Bertl i.e. /some/path/to/file 1115512781 M * Bertl will cause recursive lookup which can be reduced to having /some/path/to and looking up file 1115512804 M * Bertl this lookup can have two results, either a dentry or a negative dentry 1115512825 M * Bertl the dentry has to resolve to the inode, unless it's a negative one 1115512839 M * Bertl (which basically states that this file does not exist) 1115512895 M * Bertl (let me know if that still leaves some questions ;) 1115512921 M * Doener` no, the source should tell me the rest :) thanks 1115512940 M * Bertl you're welcome! 1115514026 M * Bertl btw, nobody tested vserver for 'um' arch for some time now, no? 1115514037 J * explasm__ ~explasm@p549FDF88.dip.t-dialin.net 1115514048 M * Bertl welcome explasm__! 1115514134 Q * eXplasm2 Read error: Operation timed out 1115514475 M * Doener` at least i didn't test on um... ;) 1115517065 M * Bertl hmm, guess I'm off to bed now ... feeling a little tired ... 1115517079 M * Bertl have a nice whatever everyone! 1115517084 N * Bertl Bertl_zZ 1115517122 M * Doener` night Bertl_zZ! 1115520035 Q * Jani Ping timeout: 480 seconds 1115520868 N * Doener` Doener|gone 1115524554 N * Doener|gone Doener 1115529080 Q * ntrs Ping timeout: 480 seconds 1115529581 N * lilo_ lilo 1115543957 Q * albeiro Read error: Operation timed out 1115544098 J * albeiro albeiro@albeiro.usercloak.oftc.net 1115546996 J * sebd ~sebd@lesdeveloppementsdurables.org 1115548624 M * aba Hm, can somebody please give me hints howto setup proc security? I have the problem that I can't really lookup the web as long as that vserver is not running :( 1115550979 J * ntrs ntrs@Dardeene-68.188.50.87.charter-stl.com 1115552972 Q * ntrs Read error: Connection reset by peer 1115552981 J * ntrs ruzin@Dardeene-68.188.50.87.charter-stl.com 1115554533 J * Jani ~Jani@G94a9.g.pppool.de 1115554535 M * Jani Hi all 1115554598 M * aba ok, I got it up enough by myself now ... 1115557309 M * yarihm which is the patch that works against 2.6.11.8 again? 1115557424 M * yarihm ah, things that didn't work are rather trivial to merge i just see 1115557923 M * daniel_hozac 2.0-pre4 is against 2.6.11.8. 1115558115 J * prae ~prae@sherpadown.net 1115558164 N * prae prae|aw 1115558407 Q * explasm__ Quit: Verlassend 1115558426 J * eXplasm ~explasm@p549FDF88.dip.t-dialin.net 1115559850 Q * DaCa Ping timeout: 480 seconds 1115563284 Q * SiD3WiNDR Ping timeout: 480 seconds 1115563499 J * SiD3WiNDR luser@bastard-operator.from-hell.be 1115565400 J * Shuri sjnesjd@64.235.209.226 1115566772 J * DaCa ~danny@mail.limehouse.org 1115567316 Q * Jani Ping timeout: 480 seconds 1115568614 N * Bertl_zZ Bertl 1115568618 M * Bertl morning folks! 1115568787 M * eyck 3morning 1115568915 M * Bertl hey three at once! great! 1115569338 M * Bertl k, off for dinner now .. back later ... 1115569346 Q * _BWare_ Read error: Operation timed out 1115569357 N * Bertl Bertl_oO 1115569824 J * _BWare_ ~bware@office.intouch.net 1115571184 J * mep_ mep@p5091F637.dip.t-dialin.net 1115571344 Q * mep Read error: Operation timed out 1115571851 Q * monrad Ping timeout: 480 seconds 1115571978 N * Bertl_oO Bertl 1115571991 M * Bertl back now ... 1115572220 Q * Vudumen Ping timeout: 480 seconds 1115572439 J * monrad ~monrad@213083190130.sonofon.dk 1115572449 J * Vudumen vudumen@perverz.hu 1115572511 M * Bertl welcome monrad! Vudumen! 1115572737 Q * eXplasm Quit: Verlassend 1115572822 J * eXplasm ~explasm@p549FDF88.dip.t-dialin.net 1115573158 M * Vudumen hi Bertl 1115577128 M * Bertl k, back later ... 1115577132 N * Bertl Bertl_oO 1115577830 Q * sebd Remote host closed the connection 1115577836 N * BobR_oO BobR 1115579679 Q * alexx Quit: Bye 1115579718 J * alexx ~alexx@82.225.136.176 1115580469 J * kjo nobody@p5484F892.dip.t-dialin.net 1115580706 Q * steve--` Read error: Connection reset by peer 1115581027 J * ft bigfoot@194.255.113.16 1115581056 M * ft Hi, has anyone had any success using vserver on Debian PPC? 1115581150 M * ft I just can't get it to work, I've tried both testing and unstable (And thus kernel 2.6.8 and 2.6.11) with different results 1115581171 M * ft Neither working :) 1115581272 M * daniel_hozac what doesn't work? 1115581284 M * ft On unstable I get "/proc/uptime can not be accessed. Usually, this is caused by procfs-security. Please read the FAQ for more details " 1115581291 M * ft The FAQ doesn't really seem to help me, though 1115581318 M * ft And /etc/init.d/vprocunhide start writes "Fixing /proc entries visibility...ERROR" 1115581365 M * ft According to /usr/lib/util-vserver/vprocunhide, which it runs, it returns error code 3 meaning "operation failed on every file" 1115581389 M * daniel_hozac so what patch did you apply to the kernel? 1115581391 J * Doener` ~doener@p54876409.dip.t-dialin.net 1115581403 M * ft The one from Debian... kernel-patch-vserver 1115581424 M * daniel_hozac does vserver-info - SYSINFO show the right syscall for PPC? 1115581454 M * ft Let me SSH into it, makes pasting a lot easier, hang on 1115581488 M * ft volapyk:~# vserver-info 1115581488 M * ft Versions: 1115581488 M * ft Kernel: 2.6.11 1115581489 M * ft VS-API: volapyk:~# 1115581543 Q * Doener Read error: Operation timed out 1115581566 M * ft Looks pretty strange to me :) 1115581613 M * daniel_hozac are you sure the vserver patch is applied to the kernel you're running? 1115581635 M * ft Yes. The entries are in the kernel configuration and enabled 1115581640 M * micah ft: how are you applying the patch? 1115581663 M * ft /usr/src/kernel-patches/all/apply/vserver 1115581668 M * ft Running that from inside the kernel source tree 1115581694 M * micah yeah that applies it appropriately 1115581716 M * ft volapyk:~/kernel-source-2.6.11# grep -i vserver .config 1115581716 M * ft CONFIG_VSERVER=y 1115581716 M * ft # Linux VServer 1115581716 M * ft CONFIG_VSERVER_LEGACY=y 1115581716 M * ft CONFIG_VSERVER_LEGACYNET=y 1115581716 M * ft CONFIG_VSERVER_PROC_SECURE=y 1115581718 M * ft # CONFIG_VSERVER_HARDCPU is not set 1115581718 M * ft # CONFIG_VSERVER_DEBUG is not set 1115582256 M * ft So... What did I do wrong? :) 1115582947 M * ft Should I try applying the "official" patches on a clean kernel tree and see if that works? 1115582957 M * mikegrb yes 1115583113 M * ft 1.9.5, then? 1115583300 N * Bertl_oO Bertl 1115583310 M * Bertl evening folks! 1115583326 M * eyck evening 1115583354 M * Bertl ft: did you install and boot your patched kernel? 1115583382 M * ft Yes 1115583386 M * Bertl ft: ah, just read, you are on PPC, no? 1115583389 M * ft Yep 1115583394 M * Bertl ppc or ppc64? 1115583397 M * ft Regular PPC :) 1115583402 M * ft 750, 32bit :) 1115583405 M * Bertl G4 ? 1115583414 M * ft G3 1115583420 M * Bertl k 1115583432 M * Bertl let's do a few checks first ... 1115583441 M * Bertl ah, which kernel is that? 1115583459 M * ft 2.6.11 (now) 1115583461 M * ft Tried 2.6.8 too 1115583489 M * Bertl okay, would you mind building another one? 1115583521 M * ft Not at all 1115583525 M * ft Anything to make it work :) 1115583545 M * Bertl okay, then let's get the 2.6.11.8 from kernel org 1115583561 M * ft Hmm, not sure if kernels from kernel.org work, but I'll give it a try 1115583562 M * Bertl and the http://vserver.13thfloor.at/Experimental/patch-2.6.11.8-vs2.0-pre4.diff 1115583573 M * Bertl yes, the kernel org kernels work quite fine 1115583581 M * ft On my platform, I mean :) 1115583593 M * Bertl uname -a 1115583593 M * Bertl Linux anson 2.6.11.8-vs2.0-pre4 #1 Sat May 7 23:52:06 CEST 2005 ppc unknown unknown GNU/Linux 1115583700 M * ft Slowly getting there with the download 1115583803 N * BobR BobR_zZ 1115583828 M * Bertl I didn't manage to get the tools running yet (on Mandrake 9.x) but that is a different story ... 1115583875 M * ft Which tools? 1115583880 M * ft To be precise:) 1115583881 M * Bertl util-vserver ... 1115583896 M * ft Hm, aren't they kinda essential? 1115583914 M * Bertl yes, and no ... there _are_ certain ways ... 1115583950 M * ft OK, building the kernel, this will take about 20-30 minutes or so 1115583989 M * ft But well, if the utils don't work, isn't that my problem? 1115583991 M * Bertl hmm, you are on debian, right? 1115583993 M * ft Yes 1115584003 M * ft Doesn't the vprocunhide stuff use util-vserver? 1115584011 M * Bertl do you ahve a 'reduced' config for the kernel? 1115584029 M * Bertl (or do you use the 'debian default' config) 1115584053 M * Bertl ft: yes, we have to make the tools work for you ;) 1115584071 M * ft I use a modified version of the default Debian config ;) 1115584079 M * Bertl okay, good ... 1115584103 M * Bertl in the emantime, you already booted with the vserver patched 2.6.11 no? 1115584127 M * ft Yes 1115584150 M * Bertl okay, so let's do a few tests/prepartions with this one (while it is compiling) 1115584161 M * ft OK 1115584180 M * Bertl cat /proc/virtual/info 1115584219 M * ft volapyk:~# /etc/init.d/vprocunhidcat /proc/virtual/info 1115584219 M * ft -su: /etc/init.d/vprocunhidcat: No such file or directory 1115584220 M * ft Ehehe 1115584226 M * ft OPop 1115584228 M * ft Wrong window nayway 1115584231 M * ft Let's try again 1115584233 M * ft Without the typos and stuff 1115584241 M * ft volapyk:~# cat /proc/virtual/info 1115584241 M * ft VCIVersion: 0001:0025 1115584241 M * ft VCISyscall: 257 1115584241 M * ft VCIKernel: 03000016 1115584242 M * ft There 1115584249 M * Bertl good ... 1115584274 M * Bertl you might want to enable the debugging options in the new kernel ;) 1115584291 M * ft Which debugging options to be exact? :) 1115584308 M * Bertl CONFIG_VSERVER_DEBUG 1115584317 M * ft OK 1115584320 M * Bertl you won#t need the history tracing ... 1115584338 M * Bertl and the option depends on the main kernel debug IIRC 1115584346 M * ft OK, rebuilding :) 1115584363 M * Bertl btw, you do not need a make clean or so if you change something 1115584383 M * Bertl the 2.6 build system knows pretty well _what_ to rebuild 1115584426 M * ft Well, the few times I've made minor changes to my config, it has rebuilt everything every time :) 1115584432 M * Bertl ft: is this an apple or something different? 1115584436 M * ft Something different 1115584441 M * ft Called Pegasos 1115584521 M * Bertl hmm, looks nice ... 1115584573 M * Bertl I assume the Pegasos I specs apply, no? 1115584596 M * ft Yeah 1115584608 M * ft I have both (Pegasos 1 and Pegasos 2), but I'm installing this on the Pegasos 1 1115584699 M * Bertl I'm pretty confident we get this working ... 1115584718 M * ft If the kernel.org sources work on the Pegasos 1 ;) 1115584813 M * Bertl yeah, well, looks good to me ... from the specs 1115584823 M * Bertl okay, next thing .. please get this one: 1115584835 M * Bertl http://vserver.13thfloor.at/Experimental/SYSCALL/vdlimit-0.04.tar.bz2 1115584836 M * ft Well, the firmware version on the Pegasos 1 is a bit fucked up :) 1115584882 M * ft I'm not sure the patches to make it work on the Pegasos 1 made it into the main tree... But we'll see 1115584951 M * Bertl http://www.pegasosppc.com/operating_systems.php 1115584980 M * Bertl telling from the list of supported distros, I#d say they have ... 1115585016 M * Bertl what bootloader do you use? yaboot? 1115585028 M * ft Marketing fluff ;) 1115585033 M * ft Might as well just be for Pegasos 2, not Pegasos 1 1115585036 M * ft (which isn't made or sold anymore) 1115585044 M * ft No bootloader 1115585076 M * Bertl so directly booting from OF? 1115585079 M * ft Yep 1115585104 M * Bertl interesting .. never considered that on my PP 1115585111 M * Bertl *PB 1115585264 M * Bertl did you manage to download the vdlimit tar? 1115585294 M * ft Yep 1115585302 M * Bertl okay, also get this one: http://vserver.13thfloor.at/Experimental/SYSCALL/syscall.h 1115585317 M * Bertl and replace the syscall.h inside the vdlimit-0.04 tree 1115585326 M * ft k 1115585336 M * Bertl then do 'make clean && make' 1115585393 M * ft OK, built 1115585402 M * ft And so is the kernel 1115585411 M * Bertl did you get a warning from the vdlimit compile? 1115585439 M * ft Yes 1115585443 M * ft vdlimit.c:66: warning: comparison is always true due to limited range of data type 1115585450 M * Bertl yeah, just saw it ;) 1115585475 M * ft OK, should I boot the 2.6.11.8 kernel now? 1115585478 M * ft If it works:) 1115585489 M * Bertl no it doesn't ;) 1115585496 M * ft It doesn't? 1115585546 M * Bertl but this one does: http://vserver.13thfloor.at/Experimental/SYSCALL/vdlimit-0.05.tar.bz2 1115585562 M * ft Well, the kernel, I meant 1115585584 M * ft Should I still replace syscall.h? 1115585586 M * Bertl ah, yeah, let's try it 1115585594 M * Bertl (the booting I emant ;) 1115585609 M * Bertl ad vdlimit, no it already has the latest 1115585614 M * ft OK 1115585664 M * ft Holy cow. It boots. 1115585676 M * eyck Holy cow. 1115585682 M * Bertl hehe 1115585763 M * ft Well, need a food break. BBS :) 1115585775 M * Bertl enjoy your meal! 1115585845 M * ft Ta :) 1115585937 M * Bertl for the record: 1115585974 M * Bertl Linux-VServer Test [V0.12] (C) 2003-2005 H.Poetzl 1115585974 M * Bertl chcontext is working. 1115585974 M * Bertl chbind is working. 1115585974 M * Bertl Linux 2.6.11.8-vs2.0-pre4 s390x/0.30.207/0.30.207 [Ea] 1115585974 M * Bertl VCI: 0001:0025 263 03110076 1115585976 M * Bertl --- 1115585979 M * Bertl [000]# succeeded. 1115585981 M * Bertl [001]# succeeded. 1115585984 M * Bertl [011]# succeeded. 1115585986 M * Bertl [031]# succeeded. 1115585989 M * Bertl [101]# succeeded. 1115585991 M * Bertl [102]# succeeded. 1115585994 M * Bertl [201]# succeeded. 1115585996 M * Bertl [202]# succeeded. 1115587027 M * daniel_hozac nice. 1115587073 M * Bertl yeah, was quite some work ... with the hercules simulator 1115587442 M * eyck hercules is not very fast, 1115587482 M * eyck Hmm, I've got contect with some s390 folks, I told them about vserver and they were rather interested 1115587495 M * eyck hmm, that was a year ago though :( 1115587569 M * Bertl yes, but the hercules emulator works surprisingly well, once you figured how to boot it ;) 1115587641 M * eyck well, I tried booting debian and installing oracle on hercules once, and it wasn't a very pleasant experience 1115587673 M * eyck hmm, that was on rather slow hardware though, some p4 and 0.5G o' ram iirc. 1115587697 M * eyck good to hear that nothing's stopping vserver though. 1115587740 M * Bertl well, kernel compile took a day or so ... 1115587753 M * Bertl (native on the system of course ;) 1115588228 J * jebba ~jebba@c-67-176-38-246.hsd1.co.comcast.net 1115588276 M * Bertl welcome jebba! 1115588329 M * jebba hey Bertl 1115588399 M * ft OK, I'm back :) 1115588401 M * ft Sorry for the delay 1115588424 M * Bertl np! 1115588441 M * ft OK, got vdlimit-0.05 built 1115588446 M * ft And 2.6.11.8 booted 1115588478 M * Bertl excellent ... 1115588487 M * Bertl now try ./vdlimit -V 1115588498 M * ft version: 0001:0025 1115588504 M * ft I'm without network ATM, so pasting manually :) 1115588509 M * ft Let me fix the driver, hang on 1115588606 A * Bertl hanging on ... 1115588618 M * ft :) 1115588724 M * ft OK, up and running with networking :) 1115588769 M * ft So, what now? 1115588835 M * Bertl well, you have dietlibc installed 1115588850 M * ft Nope? 1115588867 M * Bertl okay, get the 0.28 release from their site (as tar) 1115588873 M * Bertl and the following patches: 1115588914 M * Bertl http://vserver.13thfloor.at/Stuff/DIETLIBC/dietlibc-gethostbyname-fix01.diff 1115588927 M * Bertl hmm, sec checking regarding syscall ... 1115589201 M * Bertl yeah, guess we have to fix that too .. 1115589243 M * ft OK? :) 1115589378 M * Bertl well, you can download the tools (util-vserver 0.30.207) in the meantime 1115589451 Q * kjo Quit: Verlassend 1115589513 M * ft OK 1115589517 M * ft Debian ones won't do, then, I guess? 1115589554 M * micah is there a patch against 2.6.11.8? I notice the one that exists on 13thfloor is against 2.6.11.5 and fails against .8 1115589557 M * Bertl well, the tools might, the dietlibc won#t its borken :/ 1115589570 M * ft I haven't been able to get install any dietlibc through Debian :) 1115589591 M * Bertl micah: http://vserver.13thfloor.at/Experimental/patch-2.6.11.8-vs2.0-pre4.diff 1115589617 M * ft OK, got http://www.13thfloor.at/~ensc/util-vserver/files/alpha/util-vserver-0.30.207.tar.bz2 1115589774 M * Bertl okay, got the dietlibc too? 1115589779 M * ft Yep 1115589786 M * ft And patched with the gethostbyname patch 1115589790 M * ft Didn't build any of them yet, though 1115589799 M * Bertl good, we ened another 'patch' 1115589810 M * Bertl that's what I'm working on right now ;) 1115589814 M * ft M'kay :) 1115589821 M * ft Should I go fetch a hammer and a saw? 1115589830 M * Bertl if it helps you ;) 1115589833 M * ft :) 1115589900 M * micah Bertl: thanks, however I was mistaken, it appears that the 2.6.11.5-vs1.9.5 patch applies to 2.6.11.8 fine, I was trying against 2.6.12-rc4 1115589919 M * Bertl yes, but you really should use the new patch ;) 1115590047 M * micah Bertl: i have been trying to find what has changed from 1.9.5 to 2.0, but it is hard unless I read many IRC logs :) 1115590101 M * Bertl no, it's easy ... sec 1115590110 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.6/ 1115590115 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-2.0/ 1115590122 M * Bertl this basically changed ... 1115590176 M * micah the directory? :) 1115590187 M * micah hehe 1115590192 M * Bertl the patches inside ;) 1115590199 M * ft Someone's got about as good a humour as I have ;) 1115590371 M * ft Anything more I can do while waiting? 1115590426 M * Bertl study the code ... if you like to ... 1115590452 M * Bertl ft: hmm, are you good at ppc assembler? 1115590473 M * ft Well, I know enough to debug my code 1115590483 M * micah Bertl: patches certainly show the actual changes made, but many would like a higher-level list of what has changed, for reading patches is hard :) 1115590488 M * ft The syntax of some of the more complex shift instructions manage to escape me all the time, though :) 1115590509 M * Bertl micah: will do when I get around, but the 'names' are very descriptive 1115590526 M * micah are the patches/changes checked into a source code repository somewhere that I could go through the changes and create a "manifest" of changes? 1115590548 M * micah Bertl: (I am offering my help, but if you prefer to do it later its ok :) 1115590571 M * Bertl andis. r9,r0,4096 1115590571 M * Bertl beqlr+ r0 1115590571 M * Bertl lis r9,errno@ha 1115590571 M * Bertl stw r3,errno@l(r9) 1115590571 M * Bertl li r3,-1 1115590574 M * Bertl blr 1115590621 M * ft Yes? 1115590630 M * Bertl does this look reasonable for checking r9 against -4096 and if nearer to zero return the value 1115590649 M * ft Not at all 1115590652 M * Bertl no if not between -4096 and 0 then return the value 1115590655 M * ft It does r9&4096 1115590664 M * Bertl sec 1115590694 M * ft If r9&4096 is 0, it jumps to whatever is in r0, else it stores r3 in errno and returns -1 from the function 1115590726 M * ft Uhm, andis 1115590753 M * Bertl from my notes, I have this: 1115590758 M * ft That'd be 4096<<16 ;) 1115590766 M * Bertl sret: r3 1115590766 M * Bertl serr: (carry) 1115590784 M * Bertl err & 0x10000000 = error 1115590792 M * Bertl that looks fine then ... 1115590815 M * ft Yep 1115590824 M * Bertl but the registers don't match my notes ... sec 1115590880 M * Bertl okay, r0 is the syscall return value 1115590894 M * Bertl so it does r0 & 0x10000000 -> r9, no? 1115590907 M * ft Yes 1115590908 M * Bertl this should change the flagword, and thusly 1115590916 M * Bertl beqlr+ r0 should return r0 1115590921 M * ft No 1115590925 M * ft It jumps to r0 1115590934 M * Bertl sure? 1115590955 M * ft It jumps to the whatever r0 points to, if r0&0x1000000 == 0, that is 1115590981 M * ft +0 1115591000 M * ft Where is that code? 1115591052 M * Bertl I'm currently building it ... from the disasm of libc and my notes 1115591109 M * ft Hmm, shouldn't that be beql, though? 1115591111 A * ft checks something 1115591132 M * ft Yes, beqlr is invalid ;) 1115591143 M * ft You probably meant beql, or beqlr with no argument 1115591153 M * ft If you want to return whatever is in r0, do mr r3,r0 then beqlr 1115591174 M * Bertl no, r0 is the default return register, no? 1115591178 M * ft No 1115591179 M * ft r3 is 1115591203 M * ft r0 might be used for return values from syscalls, dunno, but normal sysv function return in r3 ;) 1115591216 M * Bertl ah, okay, now I think I figured the code ... sec 1115591262 M * Bertl sc 1115591262 M * Bertl mfcr r0 1115591271 M * Bertl get flagword in r0 1115591275 M * Bertl andis. r9,r0,4096 1115591280 M * Bertl make the check 1115591284 M * Bertl beqlr 1115591295 M * Bertl return (with value from r3 syscall) if not set 1115591304 M * Bertl lis r9,errno@ha 1115591304 M * Bertl stw r3,errno@l(r9) 1115591312 M * Bertl store error code to errno 1115591317 M * Bertl li r3,-1 1115591323 M * Bertl load -1 into r3 1115591328 M * Bertl blr 1115591331 M * Bertl return 1115591334 M * Bertl looks fine? 1115591340 M * ft Yeah, notes match, at least :) 1115591351 M * yarihm hey everyone 1115591353 M * ft I'm not sure what's in bit 28 of cr, though, so that'll be your problem :) 1115591361 M * Bertl k, let me heck what the assembler says to that ;) 1115591379 M * Bertl evening yarihm! 1115591381 M * yarihm i've googled for my error but nothing found yet ... it is probably debian related. during boot i get lots of messages ..... 1115591386 M * yarihm yo Bertl :) 1115591398 M * yarihm so, i get lots of messages, and they look like this: 1115591402 A * yarihm runs to the terminal 1115591412 M * Bertl interesting message ;) 1115591439 M * ft :) 1115591447 M * yarihm /proc/foobar: Function not implemented 1115591472 M * yarihm it must be related to some vserver-thing that's starting up, i'll just checkout the init-script 1115591473 M * Bertl yarihm: yeah, that look debian related ... 1115591477 M * ft I think I got something similar with testing, can't remember anymore :) 1115591496 M * yarihm Bertl: could it be that it is because i turned on proc-security in the kernel-config? 1115591502 M * Bertl ft: are registers specified as %r0 or r0 or just 0 ? 1115591514 M * ft Depends on the assembler. gas wants 0 or %r0, I think 1115591515 M * ft Or %gpr0 1115591521 M * yarihm ft: you wouldn't remember how to fix it, would you? 1115591523 M * ft Either of those 3 1115591532 M * ft No, I'm using unstable now and got entirely different problems :) 1115591586 M * Bertl to be fair, you should mention the ppc arch ;) 1115591591 M * yarihm Bertl: there sure has been someone coming here wining for this, no? :) 1115591594 M * ft I'm running on PPC, BTW :P 1115591598 M * yarihm :D 1115591599 M * Bertl ft: ah, yes builds fine ... 1115591611 M * Bertl yarihm: which kernel/patches/tools 1115591632 M * yarihm Bertl: uh, wait, my bad, my most bad 1115591648 M * ft My bed! Longing for me! :) 1115591653 M * yarihm it is an unpatched kernel ... *hust* ... sorry 1115591659 M * ft :) 1115591670 M * Bertl yarihm: hmm, well, then I'd suggest to change that ;) 1115591675 M * yarihm most embarassing, once again 1115591698 M * yarihm Bertl: are you sure that might change something? i'm rather sure it is something else indeed 1115591700 M * yarihm ,)) 1115591738 M * Bertl well, you 'probably' have vprocunhide running on startup 1115591743 M * ft Hm, bit 0 of cr0 is "negative" 1115591759 M * Bertl yarihm: and this (of course) does use vserver syscall commands 1115591764 M * ft Wouldn't replacing mfcr, andir, beqlr by bltlr work? 1115591769 M * yarihm Bertl: i'm just kidding ... you are exactly right, that's what is running and produces these errors 1115591849 M * Bertl ft: http://vserver.13thfloor.at/Stuff/DIETLIBC/syscall.S 1115591866 M * Bertl get this and put it into libcompat (of your dietlibc 0.28) 1115591885 M * ft OK 1115591886 M * Bertl you can later improve the syscall code for ppc 1115591896 M * Bertl hey at least it compiles ,) 1115591920 M * ft Does it work too? :) 1115591926 M * ft OK, copied in 1115591941 M * Bertl well, can't test it .. as I said, my PB has other issues with the tools 1115591959 M * yarihm Bertl: i saw a SPARC64-page on the website ... do you think it is safe running vserver on a SPARC64? 1115591960 M * ft :) 1115591978 M * Bertl uname -a 1115591978 M * Bertl Linux ankaa 2.6.11.8-vs2.0-pre3 #7 SMP Tue May 3 02:18:13 CEST 2005 sparc64 GNU/Linux 1115591982 M * Bertl I guess so ;) 1115591988 M * yarihm :) 1115591997 M * yarihm what sort of machine do you have? 1115592004 M * ft OK, so now I build dietlibc? 1115592039 M * Bertl yup 1115592050 M * Bertl yarihm: it's not mine, but it's a sparc ultra 1115592051 M * ft Just make will do? 1115592062 M * yarihm i bought this Enterprise 3000 ... very nice machine 1115592074 Q * prae|aw Quit: Pwet 1115592077 M * Bertl ft: yup, you might want to 'configure' it though 1115592106 M * ft What is there to configure? 1115592161 M * Bertl IIRC where you isntall it 1115592180 M * ft Hm, /opt/diet? :) 1115592183 M * yarihm will the new vserver-tools be linked against fefes libc? 1115592200 M * ft Will /usr/local work? Or will it clash with something? 1115592204 M * Bertl hmm, well obviously the install is not done ;) 1115592233 M * Bertl so best do what I did, just build it and copy the diet binary into the path ;) 1115592244 M * ft Hmm, OK :) 1115592257 M * Bertl yarihm: new means? 1115592325 M * yarihm Bertl: those to come, those i've not tested yet :) i didn't notice a dependancy to dietlibc here 1115592356 M * Bertl util-vserver (alpha) uses dietlibc yes 1115592370 M * ft OK, make done 1115592390 M * ft So, what do I copy, and where? :) 1115592408 M * Bertl ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var --with-vrootdir=/vservers 1115592417 M * Bertl this is what I do to configure util-vserver 1115592433 M * Bertl copy the bin-ppc/diet into /usr/bin/ 1115592453 M * ft Done :) 1115592476 M * Bertl attention: the pathes used after this configure are _not_ debian conform 1115592498 M * Bertl but I jsut don't remember where they put the stuff ;) 1115592499 M * ft Am I supposed to run that ./configure, or is he? :) 1115592516 M * Bertl ft: no you, once you are in util-vserver-0.30.207 dir 1115592530 M * yarihm when using util-vserver 0.30.204, should i use 2.0-preX or the 1.9.5-branch? 1115592532 M * ft OK, so I'd better find the correct args first? 1115592551 M * ft What is "vrootdir"? 1115592557 M * ft Well, nevermind, hang on 1115592562 M * Bertl ft: correct? no idea, the debian ones are an option ;) 1115592563 M * yarihm ft: the place where the vservers are installed 1115592590 M * Bertl yarihm: 2.0-pre should work with older tools but is best utilized with 0.30.207++ 1115592592 M * DaCa its /var/lib/vservers on debian 1115592596 M * yarihm heck, who came to the idea to set /var/lib/vservers to be FSH-compilant? 1115592610 M * yarihm Bertl: i'll check for a backport 1115592632 M * yarihm Bertl: if I don't find one, should i go for 1.9.5? so that the utils and kernel match somewhat? 1115592666 M * Bertl well, you can if you like, but I don#t see a point in doing so if 2.0-pre4 does work too ... 1115592670 M * ft DaCa, and is that the only I need to change? 1115592722 M * DaCa ft: I am not sure about the local-state one 1115592725 M * ft Bertl, can I build the Debian version instead but with the diet I just built? 1115592824 M * Bertl the debian version of 0.30.207 probably 1115592855 M * ft Yeah, except I get lots of errors during build, oh dear 1115592869 M * Bertl what errors? 1115592871 M * ft In file included from /home/bigfoot/dietlibc-0.28/include/unistd.h:6, 1115592871 M * ft from lib/vserver-internal.h:30, 1115592871 M * ft from lib/syscall.c:33: 1115592876 M * ft /home/bigfoot/dietlibc-0.28/include/sys/stat.h:616: error: syntax error before "mode_t" 1115592879 M * ft That's the first one 1115592906 M * Bertl hmm, interesting ... 1115592997 M * Bertl ah, yes, I remember ... sec 1115593006 M * ft :) 1115593065 M * Bertl try something like find /home/bigfoot/dietlibc-0.28/include -type f -exec sed -i 's|(powerpc)|(__powerpc__)|g' {} \; 1115593082 M * ft M'kay 1115593100 M * DaCa ft: localstate is ok, but debian also adds --mandir=\$${prefix}/share/man --infodir=\$${prefix}/share/info 1115593101 M * Bertl or you could get the dietlibc source and add the two changes we did? 1115593130 M * ft You mean from Debian? 1115593143 M * Bertl yeah, well, guess they have it there ;) 1115593144 M * ft Gnar 1115593149 M * ft Hang on :) 1115593153 M * ft I keep fucking up stuff 1115593358 M * ft OK, running the find line now 1115593415 M * Bertl ah, just realized there _are_ some debian folks around today! ;) 1115593493 M * Bertl ft: if that doesn't work, then the 'patching the debian sources' approach might be better ... 1115593503 M * ft Well, it's building util-vserver now 1115593529 M * ft Is anyone gonna fix all this in Debian and/or respective projects, BTW? :) 1115593565 M * Bertl don#t look at me, I reported all issues (and will post the ppc once it works) to fefe and I keep telling the debian folks ;) 1115593575 M * ft Hehe, OK :) 1115593579 M * Bertl until now, nothing was fixed ;) 1115593582 M * ft Just so the next poor guy doesn't have to go through the same stuff :) 1115593593 M * ft Yeah, well, I reported a security bug once, took 9 months to get it fixed IIRC :) 1115593611 M * ft I know of one problem I haven't reported yet, though 1115593627 M * ft The -x flag of sox doesn't work anymore (at least on x86 :) 1115593629 M * ft I'm using an old package 1115593630 M * Bertl hmm, well, so we should ahve that in upstream in about 7 month ... 1115593633 M * ft (-x = byteswap) 1115593659 M * Bertl this was too dangerous, 'cause of infection ;) 1115593706 M * ft Hm, still building :) 1115593720 M * Bertl it takes quite some time ... many tools 1115593785 M * ft Poor G3 ;) 1115593868 M * ft OK, built 1115593880 M * Bertl k, install it and get the testme.sh 1115593910 M * ft Where is testme.sh? :) 1115593939 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh 1115593955 M * ft Should run it as root? 1115593965 M * Bertl yup 1115593986 M * ft 1115593989 M * ft volapyk:/home/bigfoot# sh testme.sh 1115593990 M * ft Linux-VServer Test [V0.12] (C) 2003-2005 H.Poetzl 1115593990 M * ft chcontext is working. 1115593990 M * ft chbind is working. 1115593990 M * ft Linux 2.6.11.8-vs2.0-pre4 ppc/0.30.207/0.30.207 [Ea] 1115593990 M * ft VCI: 0001:0025 257 03010016 1115593992 M * ft --- 1115594005 M * ft [000]# succeeded. 1115594005 M * ft [001]# succeeded. 1115594005 M * ft [011]# succeeded. 1115594005 M * ft [031]# succeeded. 1115594005 M * ft [101]# succeeded. 1115594006 M * ft [102]# succeeded. 1115594006 M * ft [201]# succeeded. 1115594009 M * ft [202]# succeeded. 1115594013 M * Bertl excellent! 1115594042 M * ft Does that mean it works? :) 1115594049 M * Bertl could you /msg me the /proc/cpu info in private? 1115594051 M * ft I get some debug output on the console 1115594053 M * ft startup/shutdown 1115594057 M * ft "returned with sync 256" 1115594068 M * Bertl that's quite fine, we'll address that in a jiffie 1115594091 M * ft Is that the debug output I compiled into the kernel? 1115594106 M * Bertl yep, I assume I may publish your success (on the ml) ... 1115594119 M * ft Sure 1115594149 M * Bertl k, /proc/cpuinfo ? 1115594160 M * ft processor : 0 1115594161 M * ft cpu : 745/755 1115594161 M * ft temperature : 46-48 C (uncalibrated) 1115594161 M * ft clock : 601MHz 1115594162 M * ft revision : 51.17 (pvr 0008 3311) 1115594162 M * ft bogomips : 1200.12 1115594164 M * ft machine : CHRP Pegasos 1115594234 M * ft Holy shit, vserver vtest start seems to be working :) 1115594250 M * Bertl of course, what did you expect ;) 1115594258 M * ft Hmm, not to work :) 1115594265 M * Bertl now before you stop anything 1115594270 M * Bertl let's do a few adjustments ;) 1115594275 M * ft M'kay :) 1115594282 M * ft Waiting for DNS timeout from "Starting MTA" :) 1115594361 M * Bertl get in your /etc/vservers dir, wherever it might be located 1115594385 M * ft Exactly there 1115594393 M * Bertl do 'mkdir .defaults/apps/vshelper-delegate 1115594415 M * ft Done 1115594418 M * Bertl 'echo '#!/bin/true' > .defaults/apps/vshelper-delegate/shutdown 1115594436 M * Bertl 'chmod +x .defaults/apps/vshelper-delegate/shutdown 1115594442 M * ft OK 1115594445 M * Bertl and the same for startup 1115594449 M * ft What does that do? 1115594466 M * Bertl it creates a dummy, so that the kernel helper call does not race 1115594483 M * Bertl it's basically a bug still in 0.30.207 1115594494 M * ft Oh :) 1115594504 M * Bertl and check with vserver-info - SYSINFO 1115594520 M * Bertl that you really modified the 'correct' dir 1115594532 M * ft What to check? 1115594548 M * Bertl cfg-Directory: /etc/vservers 1115594567 M * ft cfg-Directory: /etc/vservers 1115594571 M * Bertl good! 1115594585 M * Bertl the vshelper messages on the console should stop now, check with testme.sh 1115594597 M * Bertl s/stop/be silenced/ 1115594611 M * ft Yep, they did 1115594638 M * Bertl so, then, have fun with your linux-vserver setup ;) 1115594659 M * Bertl ah, we might do a some error checks for the syscall ... 1115594699 M * Bertl chcontext --xid 49152 true 1115594705 M * Bertl what does it return? 1115594719 M * ft volapyk:/home/bigfoot# chcontext --xid 49152 true 1115594720 M * ft vcontext: vc_ctx_migrate(): No such process 1115594735 M * Bertl okay, that's it, syscall return codes work fine too ;) 1115594753 M * ft Yay :) 1115594818 M * Bertl guess you're off to bed soon, but if you feel like doing so (tomorrow or so), you could make a debian dietlibc version for the next one who wants to do that ... 1115594874 M * ft Well, soonish 1115594885 M * ft Wanna poke a bit around with it now that we've got it working :) 1115594890 M * Bertl *G* 1115595057 J * hillct ~hillct@65.76.245.91 1115595066 M * Bertl evening hillct! 1115595089 M * hillct hey Bertl! 1115595127 M * hillct I'm fighting with non-vserver stupidness this evening 1115595219 M * hillct Trying to get reverse DNS to work properly 1115595240 M * ft Bertl has been playing PPC asm wizard tonight ;) 1115595252 M * hillct wow 1115595258 M * hillct Bertl is a god 1115595266 M * ft He makes things work, at least :) 1115595278 A * hillct prays before the altar ob Bertl 1115595298 M * hillct I can see it now 1115595309 M * hillct "Let there be light!" 1115595317 M * ft "Let there be vserver!" 1115595322 M * hillct "Damnit. Let's try that again" 1115595332 M * hillct "Let there be light!" 1115595346 M * hillct "maybe the third try is the charm." 1115595367 M * Bertl so, what is the problem with reverse DNS? 1115595380 M * hillct getting servfail errors 1115595390 Q * jebba Quit: Leaving 1115595394 M * hillct means I imagine I just have a broken zone file 1115595406 M * Bertl well, can you upload it somewhere? 1115595413 M * hillct sure 1115595420 A * hillct finds a home for it 1115595448 M * Bertl DaCa: you around? know something about jigdo? 1115595520 M * Bertl aba: around? jigdo? 1115595538 M * DaCa Bertl: I know a bit about it 1115595538 M * ft How do I delete a vserver I've created with newvserver? 1115595553 M * Bertl I tried to assemble a sparc dvd image 1115595566 M * Bertl jigdo-bin-0.7.1/jigdo-lite http://cdimage.debian.org/pub/cdimage-testing/dvd/jigdo-area/sparc/sarge-sparc-1.jigdo 1115595571 M * Shuri ft: stop the vserver 1115595579 M * Shuri rm -rf the directory 1115595580 M * hillct It really doesn't get simpler than this: http://vmaster.scoophost.com/120.14.193.66.in-addr.arpa 1115595603 M * ft That's it? Debian starts it on startup, or tries to, will it stop that too? 1115595605 M * Bertl DaCa: gives me: 1115595605 M * Bertl sed: -e expression #1, char 34: unterminated `s' command 1115595606 M * Bertl sed: -e expression #1, char 33: unterminated `s' command 1115595614 M * hillct all I'm trying to do is provide reverse resolution for the nameserver 1115595647 M * ft rm -rf fails for 3 files ;) 1115595649 M * Shuri delete de conf 1115595660 M * Shuri did you stop it first? 1115595665 M * ft Yep 1115595672 M * hillct I havn't seen one but has anyone put up an x86_64 vserver install HOWTO ? 1115595673 M * ft rm: cannot remove `jaca//dev/pts/0': Operation not permitted 1115595673 M * ft rm: cannot remove `jaca//proc/ide/drivers': Operation not permitted 1115595673 M * ft rm: `jaca//proc/ide/hda' changed dev/ino: Operation not permitted 1115595684 M * Shuri you didnt stop it.. 1115595692 M * ft Well, at least I tried to 1115595694 M * ft Let's try again 1115595699 M * DaCa Bertl: no clue, probably a bug? 1115595705 M * ft Server jaca is not running 1115595715 M * DaCa I havent used it recently actually 1115595735 M * ft Well, let's move the dirs out and reboot :) 1115595874 M * Bertl hillct: http://vserver.13thfloor.at/Stuff/rev.txt 1115595892 M * Bertl try that one, and maybe replace the 66.193.14.120. by the actual name 1115595941 M * hillct the NS record you mean? 1115595957 M * Bertl both, ns and main 1115595964 M * hillct main? 1115595966 M * hillct SOA? 1115595971 M * Bertl SOA yep 1115595985 M * hillct hmm 1115595988 A * hillct tries 1115596005 M * DaCa Bertl: I just installed jigdo-file on my sarge system, tried your commandline and it works out of the box 1115596013 M * Bertl but the issue probably was the IN before the NS ;) 1115596034 M * Bertl DaCa: okay, so it is an issue with sed on my test system 1115596045 M * Bertl DaCa: thanks a lot for checking! 1115596102 M * hillct thought IN was optional in NS records 1115596111 M * hillct we'll see... 1115596140 M * hillct still dead 1115596141 M * hillct er 1115596148 A * hillct bumps the serial 1115596155 M * Bertl ;) 1115596213 M * Bertl 120.14.193.66.in-addr.arpa. 86400 INPTRns1.scoophost.com. 1115596220 M * Bertl looks good ;) 1115596236 M * hillct oh? 1115596245 A * hillct still got servfail 1115596258 M * Bertl dig -x 66.193.14.120 @66.193.14.120 1115596288 M * hillct hmm 1115596304 M * Bertl everything else requires a working delegation from your IP provider 1115596305 M * hillct does that mean the upstream isn't picking it up> 1115596306 M * hillct ? 1115596313 M * hillct yah 1115596328 M * Bertl so if that _is_ there, then it isn't working ;) 1115596333 M * hillct I wonder if that's the problem. the guy swore he did the delegation a few days ago 1115596413 M * hillct not sure how the propogation works for in-addr.arpa. Is it timed or is the designated server hit until a galid result can be cached? 1115596428 M * hillct er valid result 1115596537 Q * ntrs Quit: Leaving 1115596596 M * Bertl well, it is cached too, otherwise the TTLs wouldn't make sense 1115596630 M * hillct yah, but it shouldn't cache a servfail result 1115596666 M * Bertl that is probably true ... 1115596671 M * hillct so it should keep doing realtime queries until it works at least once after which it would be cached for a period of TTL 1115596722 M * hillct er, maybe that's configurable by the upstram server 1115596798 J * BrewmeisterJim ~jim@12-214-193-211.client.mchsi.com