1165795398 Q * zxcv Quit: I love you all!
1165795553 J * Arcticfox ~arcticfox@static243-170-124.adsl.no
1165795826 M * Bertl welcome Arcticfox!
1165796445 Q * DavidS Quit: Leaving.
1165796655 M * Arcticfox hello 
1165796662 J * Johnnie ~jdlewis@jdlewis.org
1165797014 M * Arcticfox i managed to install fedora6 guest using yum on debian host... with one cavet it seems not to install the rpm database 
1165797027 M * daniel_hozac as expected.
1165797040 M * daniel_hozac package management for rpm-based guests is external, unless you internalize it.
1165797050 M * daniel_hozac (with vserver <guest> pkgmgmt internalize, IIRC)
1165797057 M * Arcticfox aha
1165797077 M * daniel_hozac make sure to install something to install packages first though.
1165797085 M * daniel_hozac i.e. vyum <guest> -- install yum
1165797365 M * Arcticfox i get
1165797367 M * Arcticfox vcontext: execvp("yum"): No such file or directory
1165797381 M * daniel_hozac did you internalize package management?
1165797587 M * Arcticfox it tells me that internal pkg management is allready initialized
1165797622 M * daniel_hozac well, if you already internalized it, it's expected to have vyum fail if you didn't install yum inside first.
1165797636 M * Arcticfox right just my tought
1165797845 M * Arcticfox ok got it working
1165799016 M * daniel_hozac so what was the problem?
1165799070 M * Arcticfox i just did at manual install
1165799160 M * Arcticfox seems that some depencies where not met according to rpm
1165799215 M * daniel_hozac what dependencies would that be?
1165799311 M * Arcticfox i just think that the rpm database is somehow corrupt
1165799350 M * Arcticfox because some of deps it clamis are missing are there ..
1165799375 M * daniel_hozac it sounds like internalizing the database didn't work out.
1165799382 M * Arcticfox yew
1165799384 M * Arcticfox yes
1165800271 Q * mountie Ping timeout: 480 seconds
1165801176 J * theops ~webchat@125-24-162-48.adsl.totbb.net
1165801201 P * theops 
1165801248 M * Bertl daniel_hozac: okay, I have a hack prototype for the device mapping
1165801297 M * daniel_hozac okay.
1165801328 M * Bertl while it basically works, it has several oddities atm
1165801340 M * daniel_hozac such as?
1165801367 M * Bertl - only the global table is implemented atm
1165801388 M * Bertl - the remapping is currently done on the host too (that is simple to fix)
1165801415 M * Bertl - it handles char and blk devices with the same mapping
1165801436 Q * sannes Ping timeout: 480 seconds
1165801445 M * daniel_hozac so device 1 3 will give it access to both character device 1,3 and block device 1,3?
1165801450 M * Bertl - because of the inode cache, the mappings 'stick'
1165801475 M * Bertl yes, but currently it does not do any access control, just 'mapping' one device to another
1165801488 M * daniel_hozac okay.
1165801509 M * Bertl nevertheless, I think it is usable for limited testing
1165801532 M * Bertl ah, I forgot another speciality, the default fallback is a 1:1 mapping atm
1165801552 M * Bertl what I figured so far is the following:
1165801563 M * Bertl - we forgot about the char vs block case :)
1165801604 M * Bertl - device inodes seem to hang around longer than usual
1165801653 M * Bertl I'll upload the patch shortly, including the debug printk's
1165801659 M * daniel_hozac ok.
1165801663 M * daniel_hozac i'll give it a spin.
1165802227 J * sannes ace@har.sagt.no
1165802348 M * Bertl daniel_hozac: http://vserver.13thfloor.at/Experimental/delta-dmap-feat0{2,3}.diff
1165802377 M * Bertl ah, yes, I forgot, you can only add new mappings for now
1165802391 M * Bertl (i.e. no replace/change/delete is working)
1165802403 M * daniel_hozac ok.
1165802465 M * Bertl interesting detail, mapping /dev/hdc to /dev/zero gives ENXIO when opened :)
1165802482 M * daniel_hozac that seems wrong...
1165802496 M * Bertl I assume it is because of the partitioned/whole disk vs 'normal' block device
1165802522 M * Bertl the mapping is done properly, the do_open() fails
1165802524 M * daniel_hozac hmm, or the fact that hdc is a block and zero is a character device?
1165802552 M * Bertl *hrmpf* okay ...
1165802566 M * Bertl anybody got a brown paperbag?
1165802577 M * daniel_hozac hehe.
1165802622 M * daniel_hozac should we put the block/character device info into the flags?
1165802643 M * Bertl not sure yes, I had that thought, see feat03
1165802645 M * daniel_hozac or just add another field to vs_mapping.
1165802652 M * daniel_hozac ah, i'm still on feat02.
1165802656 M * Bertl s/yes/yet/
1165802667 M * Bertl basically the kernel can tell that from the arguments
1165802688 M * Bertl i.e. it could have warned me in the /dev/zero case :)
1165802717 M * daniel_hozac hehe.
1165802726 M * daniel_hozac yeah, i guess i meant on the kernel side.
1165802730 M * Bertl so I think it is not really required to 'tell' what you want
1165802748 M * Bertl for the hash, I'm leaning towards using some flags there
1165802771 M * Bertl so that only specific entries are picked
1165802792 M * Bertl (especially as the hash has to be part of the vx_info too)
1165803057 M * Bertl btw, it's not completely ready/updated, but maybe it helps: http://vserver.13thfloor.at/Experimental/TOOLS/vcmd-0.07-pre1.tar.bz2
1165803083 M * raw vserver arch start; -> "No command given; use '--help' for more information. an error occured whule executing the vserver startup sequence; when there are no other messages, it is very likely that the init-sctipt () faild.
1165803127 M * daniel_hozac vdevmap seems to be able to add them.
1165803134 M * Bertl excellent!
1165803145 M * daniel_hozac raw: sysv initstyle and no /etc/{init,rc}.d/rc script?
1165803151 M * Bertl raw: script is executable? interpreter too?
1165803192 M * Bertl daniel_hozac: btw, I just saw, the chrdev mapping doesn't work here
1165803197 M * raw i'll check :)
1165803257 M * daniel_hozac oh yeah, how come?
1165803265 M * Bertl no idea, checking now
1165803740 M * Bertl strange, the remapping happens as expected
1165803751 M * Bertl nevertheless, it seems that I get the 'original' cdev
1165803825 J * Piet hiddenserv@tor.noreply.org
1165803919 M * Bertl no idea ... but I guess I'm just tired ...
1165803943 M * daniel_hozac i'm getting ret=1
1165803974 M * Bertl yeah, that is confusing
1165803978 M * daniel_hozac that means match, correct?
1165803982 M * Bertl yes
1165803986 M * raw hm i installed a debian guest... it also starts.. but if iam connection via ssh to the guest iam on the host system. any idea whats wroong?
1165803986 M * daniel_hozac okay.
1165803994 M * Bertl it is the return value of __lookup_mapping()
1165804003 M * daniel_hozac raw: your guest's sshd binds to all addresses.
1165804008 M * doener raw: you probably forgot to limit your host's sshd
1165804027 M * doener daniel_hozac: hu?
1165804035 M * raw hm
1165804036 M * daniel_hozac umm, yeah, what doener said :)
1165804046 M * Bertl daniel_hozac: should be changed to mapped=%d :)
1165804049 M * doener raw: the host's sshd may not listen on 0.0.0.0
1165804054 M * raw i that's a idea :)
1165804061 M * daniel_hozac Bertl: yeah, that'd be clearer.
1165804093 M * Bertl okay, nevertheless, I'm off for today ...
1165804105 M * Bertl i.e. have a nice one everyone! cya tomorrow!
1165804107 M * doener good night Bertl 
1165804112 N * Bertl Bertl_zZ
1165804213 J * Aiken_ ~james@tooax7-094.dialup.optusnet.com.au
1165804543 Q * Aiken Ping timeout: 480 seconds
1165804928 Q * DreamerC_ Quit: leaving
1165804946 J * DreamerC ~dreamerc@61-217-226-202.dynamic.hinet.net
1165805536 J * mountie ~mountie@CPE0080c6fe323f-CM000a739acaa4.cpe.net.cable.rogers.com
1165806226 M * raw huh! i managed to get archlinux run with sysv like initscript-calls :)
1165806242 M * raw (i just generated a simple rc script)
1165806283 Q * mountie Ping timeout: 480 seconds
1165806880 Q * Piet Quit: Piet
1165807450 M * Arcticfox daniel_hozac: i reinstalled worked 100% added yum/rpm to the "base" and the depencies got worked up
1165807806 M * raw http://linux-vserver.org/util-vserver:Howto_virtualize_an_exisiting_Linux_server <- why should sshd_config -> UsePAM be disabled on guest system?
1165808601 Q * anonc Ping timeout: 480 seconds
1165809861 M * hardwire for crazy sake
1165813819 A * micah just built first vserver with the rsync build method, very nice! (borat voice)
1165818387 M * nebuchadnezzar I have a strange message when util-vserver is started on a debian sid with 2.6.19-vs2.2.0-rc4
1165818397 M * nebuchadnezzar Fixing visibility of /proc entries for Linux-VServer guests...Kernel unaligned access at TPC[4b844c] filldir64+0x54/0x134
1165818466 M * nebuchadnezzar I past all the lignes here http://paste.linux-vserver.org/758
1165818490 M * nebuchadnezzar It runs on UltraSparc III+
1165818609 M * nebuchadnezzar It happens each time I run /etc/init.d/util-vserver restart
1165818638 M * nebuchadnezzar off for work now
1165818640 M * nebuchadnezzar see you
1165820057 Q * sannes Read error: Operation timed out
1165820289 M * sladen micah: "great succsesss!"
1165821273 J * sannes ace@har.sagt.no
1165822522 J * Aiken__ ~james@tooax6-077.dialup.optusnet.com.au
1165822848 Q * Aiken_ Ping timeout: 480 seconds
1165824147 Q * mrrm Remote host closed the connection
1165824162 J * mrrm ~urkel@tor-irc.dnsbl.oftc.net
1165824578 J * FireEgl ~FireEgl@adsl-61-147-76.bhm.bellsouth.net
1165826441 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net
1165827524 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl
1165827691 Q * _cob Ping timeout: 480 seconds
1165829737 Q * shedi Quit: Leaving
1165830176 J * bonbons ~bonbons@83.222.39.117
1165830534 N * Bertl_zZ Bertl
1165830538 M * Bertl morning folks!
1165830547 M * waldi good morning Bertl 
1165830577 M * Bertl hey waldi! already looked into 2.6.19/vs2.2.0?
1165830617 M * bonbons morning Bertl!
1165830622 M * waldi bonbons: not really
1165830633 M * waldi err, s,bonbons,Bertl,
1165830655 M * Bertl just curious, how will debian 'advance' there?
1165830683 M * Bertl hey bonbons! how are you?
1165830718 M * bonbons fine, having my week-end today :)
1165830838 M * waldi Bertl: we have too less manpower for the kernels, i'm overloaded with studies
1165830870 J * dna ~naucki@202-205-dsl.kielnet.net
1165830897 M * Bertl waldi: i.c. so debian will stick with 2.6.18 for some time?
1165830907 M * waldi Bertl: etch will release with .18
1165830953 M * waldi we have a .19 development tree, but noone add vserver to them yet
1165831140 M * Bertl well, patches are there (vs2.2.0-rc*)
1165831639 J * borgfish ~bla@141.12.9.118
1165831643 M * borgfish morning
1165831646 M * Bertl welcome borgfish!
1165831653 M * borgfish :)
1165831980 M * nebuchadnezzar hi
1165832080 Q * Borg- Quit: I CHUJ
1165832100 J * FrIkI ~FrIkI@173.Red-80-35-232.staticIP.rima-tde.net
1165832112 M * Bertl welcome FrIkI!
1165832126 J * r87hq9eer9xx ~ident_r87@173.Red-80-35-232.staticIP.rima-tde.net
1165832135 M * Bertl hey nebuchadnezzar! do you have the sparc kernel at hand, and a few minutes?
1165832135 M * r87hq9eer9xx  FrIkI es l'AMO!
1165832151 P * r87hq9eer9xx 
1165832192 M * Bertl FrIkI: hmm, what was that?
1165832206 M * FrIkI i'm sorry :p
1165832210 Q * h01ger Quit: h01ger
1165832245 M * FrIkI hi bertl
1165832333 M * raw is there a chance to keep the host from browsing the guests files?
1165832335 M * nebuchadnezzar Bertl: just IRC connexion for now
1165832358 P * FrIkI Bona nit, si et colgues!!
1165832387 J * h01ger ~holger@socket.layer-acht.org
1165832391 J * shedi ~siggi@dsl-149-109-85.hive.is
1165832402 M * Bertl raw: not really, you can make it harder for the a host admin, if you control the kernel
1165832472 M * raw the only *powerful* thing can be to start another vserver inside the vserver using qemu or something.. is this possible?
1165832477 M * nebuchadnezzar Bertl: are you ok near 17h UTC ?
1165832522 M * nebuchadnezzar Bertl: what the topic about 2.3.0.3 ?
1165832557 M * Bertl nebuchadnezzar: yeah, is fine ...
1165832579 M * Bertl raw: yeah, but ultimately won't help you either
1165832622 M * Bertl raw: i.e. the host admin can mess with your qemu binaries and/or encryption stuff and so capture your passphrases etc
1165832663 M * raw when using dm-crypt inside the qemu?
1165832685 M * raw okay the admin can backdoor my bash
1165832699 M * Bertl precisely, or capture keystrokes ...
1165832731 M * raw but it's a difference between just browsing files and starting modifiyng and backdooring
1165832732 M * Bertl raw: if you want to make sure, get a machine and lock it in your basement (behind 10" steel)
1165832757 M * raw yes :)
1165832764 M * Bertl raw: yes, you can easily use pgp or similar to protect sensitive data
1165832786 M * Bertl raw: if the host admin is interested in guest privacy, there are also some options to support that
1165832844 M * raw :) okay
1165833001 M * Arcticfox hum
1165833034 M * Arcticfox is there a way to find out who many vserver guest a dists kernel+utils support
1165833149 M * Bertl yeah, testing ...
1165833178 M * Arcticfox hehe
1165833188 M * Bertl the upper limit is currently between 49150 and 65533 depending on the config
1165833189 J * Borg- borg@aprogas.student.utwente.nl
1165833195 M * Arcticfox aha
1165833207 M * Bertl but I doubt that you can reach that limit with conventional hardware
1165833211 M * Arcticfox :D
1165833221 M * Arcticfox ok i tought max was 64
1165833229 M * Arcticfox maybe i read some old docs ..
1165833248 M * Bertl that probably was referring to the number of IPs for a guest or so
1165833258 M * Arcticfox aha ok..
1165833269 M * Bertl (which actually has a limit of 16 atm, but with patches up to 64 or so)
1165833297 M * Bertl (also note, this limit is going away ...)
1165833300 M * Arcticfox i been browsing docs + mailing lists ...
1165833318 M * Arcticfox kinda hard to melt the info sometimes
1165833319 M * Arcticfox :D
1165833323 M * Arcticfox thats great ..
1165833324 M * Bertl I know for sure, that Lycos tested with up to 250 guests
1165833353 M * Bertl (on a dual P3 system IIRC)
1165833363 M * Arcticfox hehe sounds like my setup
1165833368 M * Arcticfox well dual celeron..
1165833383 M * Arcticfox but it works really great ..
1165833399 M * Bertl it really depends on the guests (workload) and the hardware (memory/IO/network)
1165833462 Q * hardwire Ping timeout: 480 seconds
1165833464 M * Bertl how much guests you can run on without degradation in the services
1165833500 M * Arcticfox i can tell u in a few weeks when we are going to launch it into a production enviroment
1165833547 M * Arcticfox but from some simple io test ive preformed it scales very well.. 
1165833558 M * Bertl was more a dual line statement than a question actually :)
1165833569 M * Arcticfox aha :D 
1165833576 M * Bertl but of course we _are_ interested in feedback
1165833584 M * Arcticfox of course ..
1165833690 J * hardwire ~hardwire@rdbck-7535.palmer.mtaonline.net
1165833834 M * Arcticfox btw what did lycos run on those guest ?
1165833839 M * Arcticfox database  web ?
1165833848 M * Bertl no, vps for customers
1165833855 M * Arcticfox oh i see ..
1165833888 M * Bertl they are using it for some years now ...
1165834036 M * Arcticfox im using vlan in the implementation ..
1165834060 M * Bertl for the guests?
1165834063 M * Arcticfox yes
1165834075 M * Bertl yeah, that is a common one actually
1165834086 M * Bertl do you extend the vlans to the switch?
1165834092 M * Bertl (i.e. tagged vlan)
1165834125 M * Arcticfox no not atm .. 
1165834208 M * Arcticfox the cisco router we had here overheated :D
1165834213 Q * hardwire Ping timeout: 480 seconds
1165834216 M * Arcticfox so we replaced it with a obsd box
1165834280 M * Arcticfox but i let you know how the project turns out ..
1165834337 M * Arcticfox our customer is the city/county ... and they are not really opensource friendly .. but we hope to change that ..
1165834340 J * hardwire ~hardwire@rdbck-7535.palmer.mtaonline.net
1165834483 M * Arcticfox they could not believe that you could host 10-15 virtual hosts on PIII with 1G ram ... i guess people are too used to a certain os.. 
1165834606 M * Bertl definitely ..
1165834659 J * lilalinux ~plasma@80.69.41.2
1165834755 M * nox Bertl: will kvm in 2.6.20 will bring any advance for "us"?
1165834769 M * nox my stuopid question of the day (;
1165834793 M * Bertl hehe, nope, will not
1165834812 M * Bertl we are already running at full speed (100%+)
1165834830 M * nox hehe
1165834845 M * Bertl but it will help for Xen and VMware like
1165834871 M * nox thought maybe there would be some side effects
1165834910 M * Bertl I hope not (i.e. that it will get slower :)
1165834921 M * nox lol
1165835022 M * nox Bertl: you are our hypervisor (;
1165835196 M * Bertl lol, tx :)
1165835228 M * Arcticfox Bertl: you never know reg slower
1165835254 M * Arcticfox i think that for instance ext3 seems to deteriorate with every new kernel release
1165835273 Q * hardwire Ping timeout: 480 seconds
1165835348 J * mountie ~mountie@CPE0080c6fe323f-CM000a739acaa4.cpe.net.cable.rogers.com
1165835355 M * Bertl wb mountie!
1165835378 M * Bertl Arcticfox: well, constant benchmarking should show such degradations
1165835399 M * Arcticfox no i meant in general usage
1165835411 M * Bertl ah, so quality wise?
1165835415 M * Arcticfox yes
1165835533 M * nox in which point(s)?
1165835640 J * hardwire ~hardwire@rdbck-7535.palmer.mtaonline.net
1165836127 M * h01ger daniel_hozac, i finally have the patched kernel running. how do i add VXC_SECURE_MKNOD? adding it to ccapabilities doesnt work...
1165836640 Q * mrrm Remote host closed the connection
1165836646 J * mrrm ~urkel@tor-irc.dnsbl.oftc.net
1165837548 Q * Aiken__ Ping timeout: 480 seconds
1165837682 M * waldi hmm, does anyone know where I can get informations about performance impact of PAE?
1165837746 M * Bertl IIRC, for Xen it was significant
1165837785 M * waldi hmm, do you know how I can test that?
1165837856 M * Bertl http://hint.byu.edu/user.html
1165837876 M * Bertl that was used by Andrea and Ingo for discussing 4/4 splits, IIRC
1165837890 M * h01ger Bertl, hi. how do i add VXC_ capabilities? STFU didnt help...
1165837921 M * Bertl h01ger: just add it to the config file, if the tools support the capability name, it will be added
1165837936 M * Bertl h01ger: note, ccaps != bcaps
1165838066 M * Arcticfox hum
1165838078 M * waldi Bertl: lets try
1165838105 M * matti Bertl: Hello.
1165838108 M * matti Bertl: How are you?
1165838142 M * Arcticfox Bertl: is it normal that you should see the routes/dests to the host and other guests when issuing netstat -tr 
1165838200 M * Bertl nope, what kernel/patch do you use atm?
1165838204 M * h01ger Bertl,  "echo VXC_SECURE_MKNOD > bcapabilities" didnt do the trick - which config file do you mean?
1165838217 M * Arcticfox Bertl: none running stock debian
1165838236 M * Bertl h01ger: VX_C_ means ccapability :)
1165838245 M * Arcticfox maybe its b0rken(tm) 
1165838261 M * Bertl Arcticfox: could be ...
1165838269 M * h01ger "echo VXC_SECURE_MKNOD > ccapabilities" didnt work either
1165838278 M * h01ger so daniel_hozac patch is borken..
1165838341 M * Bertl do you use tools which support that capability?
1165838351 M * Bertl I mean, the tools need to be patched too
1165838361 M * h01ger uh no.
1165838372 M * Bertl alternatively, you can specify it as ^<num>
1165838374 M * h01ger daniel_hozac just wrote that patch for me..
1165838383 M * Bertl e.g. ^27 if it is bit 27
1165838402 M * Bertl that works will almost all tools
1165838455 M * h01ger so "echo ^27 >> ccapabilities"  ?
1165838462 M * Bertl for example
1165838464 M * h01ger if thats the right bit
1165838470 M * Bertl yep
1165838475 Q * ensc Killed (NickServ (GHOST command used by ensc_))
1165838485 J * ensc ~irc-ensc@p54B4ECF0.dip.t-dialin.net
1165838508 M * h01ger #define VXC_SECURE_MKNOD       0x01000000 -> so 64 or 7?
1165838522 M * Bertl Arcticfox: it would not hurt to try with a mainline kernel + patches
1165838552 M * Bertl h01ger: you have bc installed?
1165838559 M * h01ger yes
1165838572 M * h01ger somewhere at least :)
1165838581 M * Bertl use 'obase=16' and '2^7' or '2^64'
1165838662 M * kir Bertl, am I got it right that if you give a guest a permission to use mknod, you can create and access, say, /dev/sda?
1165838665 A * h01ger has to admit he is too stupid :(  echo 2^7 > ccap..
1165838677 M * h01ger ?
1165838678 M * kir Bertl, or any other device, for that matter?
1165838685 M * h01ger kir, yes
1165838717 M * Arcticfox Bertl: btw maybe its because of some of the settings in bcapabilties 
1165838724 M * Bertl kir: depends on the kernel patch, but with the default releases, yes
1165838730 M * kir h01ger, ughm. and why do you need mknod?
1165838735 M * Arcticfox Bertl: yes i was thinking doing that 
1165838737 M * h01ger thats what i try to workaround/avoid - using a proof-of-concept patch (delta-secdev-poc03.diff) from daniel_hozac 
1165838769 M * h01ger kir, to setup chroots and to create /dev/random in there, without giving the vserver the capabilities to create /dev/hda and killing themachine
1165838778 M * Bertl h01ger: it looks like 2^24 to me so ^24 :)
1165838788 A * h01ger tries - thanks Bertl 
1165838857 M * kir h01ger, looks like this problem is non-existent in openvz -- you can create any devices by default, but you can not access them unless granted from the host system
1165838920 M * kir Linux-VServer will eventually do the same I suppose
1165838926 M * Bertl kir: we have similar in experimental
1165838962 J * odim ~dim@swsoft-mipt-nat.sw.ru
1165838968 M * kir great
1165838969 M * Bertl welcome odim!
1165839083 M * Hollow ah, nice to hear that we can use mknod now..
1165839087 M * ruskie hmmm is it possible to do some of the capabilites limiting on the main system not a vserver? i.e. limit any changes to /dev ?
1165839097 M * Bertl yes
1165839108 M * Hollow this solves another problem with baselayout i guess (at least with installing it)
1165839116 M * Bertl ruskie: it is part of the Linux capability system
1165839149 M * Bertl Hollow: we are currently toying with a generic device remapping
1165839156 M * ruskie Bertl, so I could limit it to not be able to change/remove/create nodes in /dev without some extra work?
1165839173 M * Bertl yes, even on mainstream linux
1165839180 M * Hollow Bertl: sounds nice, i'm a bit behind with development currently ..
1165839183 M * ruskie hmm any docs you could point me to?
1165839232 M * Bertl ruskie: well, the Linux-VServer Paper covers the capability system somewhat
1165839266 M * Bertl but there was an FAQ too
1165839280 M * h01ger Bertl, worked now. (starting the vserver). though it took quite some time (15secs) until i saw the first service started...
1165839297 M * kir that's funny, looks like openvz is targeted to be more complete guest implementation. devices, networking...
1165839305 M * Bertl h01ger: probably some kind of udev
1165839313 M * kir but you are catching up :)
1165839333 M * Bertl kir: well, maybe you folks will catch up too :)
1165839354 M * h01ger Bertl, udev is always to blame :) and mostly rightfully :)
1165839657 M * raw  http://linux-vserver.org/util-vserver:Howto_virtualize_an_exisiting_Linux_server <- why should sshd_config -> UsePAM be disabled on guest system?
1165839689 M * Bertl no idea ...
1165839733 M * Bertl I guess it's just what the person virtualizing it did
1165839744 M * Arcticfox yeah
1165839794 Q * cdrx Ping timeout: 480 seconds
1165841385 J * DavidS ~david@vpn.uni-ak.ac.at
1165841393 M * Bertl wb DavidS!
1165841414 M * DavidS ho Bertl!
1165841964 Q * FireEgl Quit: ...
1165842028 J * cdrx ~legoater@cimai.net4.nerim.net
1165842295 M * Bertl wb cdrx!
1165842483 M * cdrx hi Bertl !
1165842547 Q * kir Remote host closed the connection
1165843731 Q * Borg- Quit: I CHUJ
1165844281 M * h01ger daniel_hozac, Bertl: yay, the patch works :)
1165844323 M * Bertl good
1165844348 M * Bertl h01ger: but don't expect too much, the feature will stay, but the interface will be different
1165844349 M * h01ger indeed. now my live-cd build is stuck at another problem, but hey :)
1165844353 M * h01ger Bertl, i know
1165844402 M * h01ger but i run sarge (or etch once its stable), so i guess i will stick with that patch for a while i guess...
1165844418 A * h01ger hugs (#)vserver :)
1165844924 J * Borg- borg@aprogas.student.utwente.nl
1165844963 A * Wonka tries to patch vs 2.2.0-rc4 into 2.6.19-git17
1165845055 M * Bertl let me know how that goes
1165845185 M * Bertl at least the pid namespace will be interesting ...
1165845458 M * Wonka several hunks do not apply. most i looked at yet because of f_dentry -> f_path.dentry (or what it was)
1165845542 M * Wonka two up to now couldn't be applied cause they already were.
1165845577 M * Bertl yeah, probably the banner fixes
1165845627 M * Wonka four...
1165845633 M * Wonka yes, some at least
1165845720 M * Wonka "child_reaper;" had gotten "child_reaper(father);"...
1165846931 J * Zaki_ ~Zaki@88.213.58.243
1165847193 M * Bertl wb Zaki_!
1165847299 Q * Zaki Ping timeout: 480 seconds
1165847533 Q * shedi Quit: Leaving
1165848928 M * Bertl okay ... nap attack ... back later!
1165848934 N * Bertl Bertl_zZ
1165849024 J * Medivh- ~ck@dslb-084-058-247-007.pools.arcor-ip.net
1165849346 J * shedi ~siggi@inferno.lhi.is
1165850105 M * Wonka arch/um/drivers/net_kern.c:92:46: error: macro "DECLARE_WORK" passed 3 arguments, but takes just 2
1165850114 Q * dna Quit: Verlassend
1165850126 M * Wonka in linux-2.6.19-git17+vs2.2.0-rc4
1165850324 M * daniel_hozac doesn't seem like something we caused :)
1165850590 J * dna ~naucki@202-205-dsl.kielnet.net
1165850637 M * Medivh- daniel_hozac, any news yet on the kernel rpms? ;) or could you maybe lemme know what patch to use? then i could build locally here
1165850652 M * daniel_hozac i'm just now committing the FC6 stuff.
1165850661 M * Medivh- ah cool :)
1165850680 M * daniel_hozac so it'll be a few hours before the builds are done.
1165850760 M * Medivh- what is upgraded in the rpms though, newer vserver patch?
1165850801 M * daniel_hozac well, everything.
1165850810 M * daniel_hozac it's against a newer Fedora kernel too.
1165850817 M * daniel_hozac http://cvs.hozac.com/viewvc/rpms/kernel/fedora-6/ has the files.
1165850862 M * Medivh- ah i see... probably the newer vserver patch which fixes my probs anyways, right?
1165850900 M * daniel_hozac possibly, though i can't remember us fixing anything like that.
1165851305 J * sebastian ~info@p54A9659A.dip.t-dialin.net
1165851390 M * h01ger daniel_hozac, dunno if you read backlog, so again: your delta-secdev-poc03.patch works fine here, thanks!
1165851403 M * daniel_hozac h01ger: yeah, i always read the backlog. ;)
1165851408 M * h01ger :)
1165851426 M * h01ger then i can leave here now, to get a little bit less distracted :)
1165851430 M * daniel_hozac that's great though, i guess it'll be useful in the device mapping feature we're working on now.
1165851434 M * daniel_hozac hehe.
1165851436 A * h01ger nods
1165851449 P * h01ger cu & thanks for the fish!
1165851564 J * ybanafa ~ybanafa@82.114.189.6
1165851755 M * ybanafa Hi All .. I installed bind9-nocap on sarge3.1 How can secure it ? 
1165851828 M * daniel_hozac what do you mean?
1165851929 M * ybanafa there is doc how to install bind9 linux-vserver & end of the doc say "secure it"
1165851981 M * ybanafa linux-vserver.org
1165852336 M * ybanafa daniel_hozac : here the link >> http://linux-vserver.org/Special:Search?search=bind9&go.x=0&go.y=0&go=Go
1165852471 M * daniel_hozac and which of those docs did you use?
1165852478 M * daniel_hozac because i don't see "secure it" in either of them.
1165852561 M * ybanafa sorry ..the one says ( you have to take care of security holes by yourself now!)
1165852585 M * daniel_hozac yes... that's to be expected whenever you stop using the distribution packages.
1165852648 M * ybanafa how can secure my vps ? thanks an advance for your help
1165852721 M * daniel_hozac again, what do you mean?
1165852767 M * ybanafa did I have to do any bind9-setting to make secure ?
1165852851 M * daniel_hozac no.
1165852861 M * daniel_hozac you'll just have to make sure you update it whenever there's a security issue.
1165852871 M * ybanafa thanks :)
1165852881 M * daniel_hozac (as always, only now it takes more effort as you have to rebuild)
1165852959 M * nebuchadnezzar back
1165852968 M * daniel_hozac Bertl_zZ: what character devices were you testing with? it looks like the drivers/char/mem.c-powered ones (/dev/[nf]ull, /dev/zero, etc.) will take some more work to get working...
1165853016 M * daniel_hozac (it decides which functions to use based on iminor(inode))
1165853089 Q * ybanafa Quit: Leaving
1165853194 M * nebuchadnezzar Bertl_zZ: sleeping ?
1165853272 M * daniel_hozac yeah, nap attack ;)
1165853436 M * nebuchadnezzar ok, do you wanna play with UltraSparc ? :-)
1165853536 M * daniel_hozac what kernel does it have?
1165853632 M * nebuchadnezzar 2.6.19-vs2.2.0-rc4
1165853661 M * nebuchadnezzar I have this error messages when I /etc/init.d/util-vserver restart : http://paste.linux-vserver.org/758
1165853697 M * daniel_hozac ah, right.
1165853729 M * daniel_hozac i take it you don't get that with vanilla?
1165853778 M * daniel_hozac but IIRC, this was a harmless warning.
1165853792 M * nebuchadnezzar I don't know what it is
1165853938 M * daniel_hozac doesn't look like that to me though... hmm.
1165853953 M * daniel_hozac do you have the vmlinux and System.map around?
1165853966 M * nebuchadnezzar yes
1165853991 M * daniel_hozac ok, do you have debugging info in the vmlinux?
1165854056 M * nebuchadnezzar only VSERVER_DEBUG*
1165854057 M * daniel_hozac i guess we'd find that out soon enough... if you get filldir64's address from System.map, and add 0x54 to it, what does addr2line -e vmlinux <address> give?
1165854230 M * nebuchadnezzar readdir
1165854761 M * tamitall w
1165854837 J * stefani ~stefani@tsipoor.banerian.org
1165855011 M * nebuchadnezzar daniel_hozac: does it speack to you ?
1165855140 J * johnny ~johnny@ip70-160-117-211.hr.hr.cox.net
1165855243 M * johnny hi folks, i'm trying to setup a spam filter proxy, and i want to make postfix only listen locally , is that possible?
1165855254 M * johnny in a vserver of course
1165855403 M * nebuchadnezzar personnaly I use a 10/ subnet for the vserver virtuall hub + firewall filetering. each vserver has it's own IP in that prefix and can speak to others according to firewall rules
1165855408 M * nebuchadnezzar does it respond to your question ?
1165855454 M * johnny i'm running them on the same vserver
1165855464 M * johnny so i'm not concerned about vservers speaking to others
1165855483 M * nebuchadnezzar 2 postfix on the same vserver ?
1165855488 M * johnny no
1165855498 M * johnny 1 postfix , 1 smtp proxy
1165855514 M * johnny i've seen alot of solutions to the question online... some are contradictory
1165855520 M * nebuchadnezzar as soon as they do not use the same IP:PORT
1165855579 M * nebuchadnezzar the proxy listen on PublicIP:smtp and the MTA to localhost:smtp for exemple
1165855588 M * johnny yeah
1165855598 M * nebuchadnezzar maybe a unix socket will be faster ?
1165855599 M * johnny i want that.. but localhost doesn't work as it stands atm
1165855607 M * johnny i don't seen an option to make it listen on a socket
1165855676 M * derjohn Bertl_zZ, daniel_hozac : I anyone think about a masking system for a guest's device nodes? I mean I have a guest for asterisk who should be able to create it's own /dev/fooX , based on restrictions set my the hist admin? And furthermore: Maybe a virtualized udev moutn for the guest? Some packages tend to created devnodes druing install, to the whole install fails. ...
1165855729 M * nebuchadnezzar johnny: it's a postfix specific quetion :-)
1165855919 M * johnny the problem applies to many packages tho
1165855965 Q * DavidS Quit: Leaving.
1165855977 M * nebuchadnezzar it's not vserver specific, on a non vserver it's the same
1165856006 M * johnny hmm? i have localhost on a non vserver tho
1165856010 M * johnny that i can ping
1165856046 M * johnny i've seen other methods of creating equivalents
1165856053 M * johnny but they dn't really go into details
1165856111 M * nebuchadnezzar or just make postfix listen on another port
1165856176 M * johnny but then something else could connect to it externally right?
1165856185 M * johnny that is the reason i haven't done that
1165856187 M * nebuchadnezzar just setup a firewall
1165856212 M * nebuchadnezzar or a dummy interface
1165856218 M * johnny that's one thing i saw
1165856228 M * johnny seen about 5 different ways to handle the situation
1165856241 M * nebuchadnezzar just use the one you prefer :-)
1165856336 P * oo 
1165856347 M * johnny uhmm.. i don't know enough to make that kinda judgement.. that's why i'm trying to ask the experts 
1165856352 M * johnny :)
1165856392 M * johnny i do have a working vserver with mysql/apache/postfix/courier-imap etc..
1165856392 M * nebuchadnezzar i'm not an expert :-)
1165856421 M * johnny so i can definitely get vserver's working ... but not anything else yet :)
1165856423 M * nebuchadnezzar I like to see vserver specific services listening on specific IP
1165856442 M * nebuchadnezzar so netstat -lp is clearer for me
1165856463 M * nebuchadnezzar so I use a private IP to bind thoses services
1165856482 M * nebuchadnezzar and let public ones bind on the public IP
1165856491 M * johnny hmm.. private ip..
1165856496 M * nebuchadnezzar on a dummy interface
1165856507 M * johnny i don't know anything about dummy interfaces.. 
1165856524 M * nebuchadnezzar modprobe dummy0 :-)
1165856760 Q * odim Quit: Leaving
1165856822 M * johnny yeah i know how to add it 
1165856836 M * johnny but i don't quite understand how it works 
1165857152 M * daniel_hozac derjohn: you have no idea how timely that comment is ;)
1165857185 M * daniel_hozac derjohn: we're working on a device mapping system now.
1165857207 M * johnny what ever happened to ngnet?
1165857247 M * daniel_hozac mainline is working on the level 2 virtualization, while 2.3 is getting improvements in the level 3 isolation we're doing.
1165857331 M * daniel_hozac derjohn: http://vserver.13thfloor.at/Experimental/delta-dmap-feat02.diff http://vserver.13thfloor.at/Experimental/delta-dmap-feat03.diff seems to do block device mapping quite fine...
1165857336 M * johnny i remember bertl talking about that sometime last year
1165857345 M * daniel_hozac yep.
1165857386 M * daniel_hozac nebuchadnezzar: i didn't quite understand your response, what was the result from addr2line?
1165857546 M * derjohn daniel_hozac, random coincidence :)
1165857875 J * Piet hiddenserv@tor.noreply.org
1165857939 M * derjohn daniel_hozac, how will the maskign be done ?
1165857943 M * derjohn *masking
1165857980 M * derjohn and: will udev wotk in a guest then ?
1165858314 Q * cdrx Ping timeout: 480 seconds
1165859880 M * nebuchadnezzar daniel_hozac: -> addr2line -e vmlinux 4b8454
1165859880 M * nebuchadnezzar readdir.c:0
1165860013 M * nebuchadnezzar daniel_hozac: that's the response for all filldir64+0xYY you can see in my past
1165860494 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net
1165860622 M * daniel_hozac nebuchadnezzar: okay...
1165860711 M * daniel_hozac derjohn: you can either allow access to a device, or do remapping (i.e. make one device appear to be another).
1165860766 M * daniel_hozac derjohn: and in the allow access department, you can define whether you want it to be creatable and/or openable.
1165860809 M * daniel_hozac (creating doesn't work yet, but opening is supposed to (though it doesn't check the ACL))
1165861492 J * DavidS ~david@217.19.46.18
1165861511 M * hardwire hoohaahaaa!
1165861858 Q * michal` Ping timeout: 480 seconds
1165862125 N * Bertl_zZ Bertl
1165862145 M * Bertl evening folks!
1165862150 M * daniel_hozac evening Bertl!
1165862157 M * daniel_hozac http://people.linux-vserver.org/~dhozac/p/k/delta-dmap-feat04.diff
1165862168 M * johnny Bertl is alive
1165862216 J * michal` ~michal@www.rsbac.org
1165862404 M * Bertl daniel_hozac: nice, here are some nitpicks
1165862423 M * daniel_hozac i have quite a few nits myself, it's a hack in progress ;)
1165862431 M * Bertl - if we pass vxi to vs_map_device() we do not want to check vx_check()
1165862444 M * Bertl   (possible divergency)
1165862455 M * daniel_hozac right, i didn't realize that until just now... :)
1165862461 M * daniel_hozac (that we passed the vxi)
1165862481 M * Bertl btw, should be enough to check for (vxi)
1165862487 M * daniel_hozac yeah.
1165862536 M * Bertl - (type & flags & DATTR_REMAP) at least needs a comment :)
1165862553 M * daniel_hozac hehe
1165862587 M * Bertl what is the purpose/semantics of the !(type & flags) check?
1165862605 M * daniel_hozac the idea was to use DATTR_OPEN.
1165862614 M * daniel_hozac and DATTR_CREATE.
1165862629 M * Bertl okay, but we have to do different checks there, no?
1165862639 Q * bronson Ping timeout: 480 seconds
1165862652 M * Bertl what I mean, we probably want some vs_device_permission() too or so
1165862653 M * daniel_hozac yeah, as is the check is void.
1165862720 M * Bertl ah, you pass the nameidata for type checks, hmm
1165862747 M * daniel_hozac yeah, 6 args just didn't seem right.
1165862769 M * Bertl what about putting that into a separate inline, which
1165862773 M * daniel_hozac i guess we could just do a simple struct { dev_t; mode_t } instead.
1165862787 M * Bertl just does the checks, and returns/sets a flag value?
1165862811 M * Bertl which is called before the do_* and stops if a mismatch is there?
1165862947 M * Bertl somehow I don't like the nameidata handing around (which has a refcount)
1165862965 M * Bertl it's too easy to 'forget' that somewhere ...
1165863052 M * Bertl but I agree, we can probably get away with a single __vs_map_device() funtion and one or two #define wrappers
1165863102 M * Bertl any idea about the character device mapping?
1165863137 M * Bertl and more important, does the feat04 work in your setup, for guests?
1165863348 M * daniel_hozac i'm still testing it.
1165863367 M * daniel_hozac the -ENXIO doesn't seem to work, nor the access check (as expected).
1165863374 M * daniel_hozac did you see my comment about the character devices?
1165863406 M * daniel_hozac 2006-12-11T17:02:55 < daniel_hozac> Bertl_zZ: what character devices were you testing with? it looks like the drivers/char/mem.c-powered ones (/dev/[nf]ull, /dev/zero, etc.) will take some more work to get working...
1165863443 M * daniel_hozac (it chooses which function set to use based on iminor(inode) in the open function)
1165863570 M * Bertl ah, hmm, that probably explains it
1165863584 M * Bertl we do not change the rdev for now
1165863620 M * Bertl I would lean towards having a separate i_mdev in addition to i_rdev and use that whenever the device numbers are used
1165863633 M * Bertl would be required for the vroot indirection anyway
1165863654 M * daniel_hozac yeah, that seems like the best solution.
1165863673 M * Bertl but OTOH, if we can avoid adding another value to the inodes
1165863695 M * Bertl e.g. bdev/cdev indirection could help here too
1165863719 M * daniel_hozac ?
1165863725 M * Bertl for each 'active' inode, we have 'two' inodes
1165863734 M * Bertl the inode, and the device inode
1165863752 M * Bertl the device inode is created when the device is discovered
1165863764 M * Bertl and it is referenced by the 'special' inode
1165863787 M * Bertl so that basically should contain the 'real' rdev
1165863802 M * Bertl while we could adjust the i_rdev in the special inode
1165863821 M * daniel_hozac the special inode is the /dev/<whatever> inode, right?
1165863826 M * Bertl yep
1165863872 M * Bertl there are a few special cases we should be aware of
1165863877 M * daniel_hozac and the device inode is i_[cb]dev? or?
1165863883 M * Bertl yes
1165863919 M * Bertl if you access a guest device on the host, it will circumvent the mapping
1165863920 M * daniel_hozac so what indirection for those would we need that we don't already have?
1165863930 M * daniel_hozac yeah, i noticed that...
1165863984 M * Bertl well, checks for i_rdev, supposed to know the 'real' dev, would have to go over i_[bc]dev, while those which should get the mapped version would go via i_rdev
1165864018 M * Bertl and we would adjust the i_rdev according to the mapping
1165864033 M * Bertl (should immediately make the memory devices work)
1165864072 M * daniel_hozac ah, yeah...
1165864108 M * Bertl but it will also affect stat/getattr as is
1165864156 M * Bertl nah, probably we want open to just use the indirection
1165864167 M * Bertl and do not do any remapping in i_rdev
1165864223 M * Bertl but I guess we can leave the details here for later
1165864243 M * Bertl the primary interest is in the open permission checks and the device creation
1165864257 M * daniel_hozac indeed.
1165864263 M * Bertl which should work fine with the existing framework (plus a few checks)
1165864341 M * daniel_hozac i can't seem to get the -ENXIO check working... did i miss something there?
1165864349 M * daniel_hozac i guess it's one of the caching issues.
1165864369 M * daniel_hozac (since the mapping printk isn't outputting anything)
1165864370 M * Bertl I will check that shortly, give me a few minutes to get something to eat
1165864388 M * Bertl i.e. I will integrate your changes into a new version
1165864389 M * daniel_hozac okay, np.
1165864463 M * Bertl btw, something different, do we want the new network interface to handle lback and bcast special?
1165864476 M * Bertl IMHO, the lback makes sense, but the bcast?
1165864584 M * matti Bertl: :)
1165864585 M * daniel_hozac i've always wondered what the bcast was good for.
1165864592 M * daniel_hozac does it even do anything as-is?
1165864612 M * Bertl not sure actually :)
1165864616 M * Bertl hey matti!
1165864706 M * matti Bertl: How are you?
1165864816 M * Bertl fine, thanks, and you?
1165864877 M * Bertl daniel_hozac: another, completely different thing: was the possibility to compile util-vserver against uClibc investigated at some point? would it be possible to 'add' uClibc support?
1165864884 M * matti Bertl: Fine :)
1165864900 M * matti Bertl: First christmas without snow ;DD
1165864903 M * matti Bertl: A bit odd.
1165864904 M * matti ;]
1165864926 M * daniel_hozac Bertl: i haven't looked at it, at least. i suppose it should be possible though.
1165864969 M * Bertl I think it might be very interesting for embedded solutions, as uClibc seems to support more platforms there
1165865066 M * matti Oh well.
1165865072 M * matti Have a nice evening everyone.
1165865078 M * Bertl you too!
1165865091 M * matti Thanks.
1165865100 M * matti :)
1165865761 J * s0undt3ch_ ~s0undt3ch@81.193.57.88
1165865852 J * _dmax ~semaj@81.193.57.88
1165866134 J * lomonaco ~IceChat7@hercules.halogica.com
1165866196 Q * dmax Ping timeout: 480 seconds
1165866198 N * _dmax dmax
1165866216 Q * s0undt3ch Ping timeout: 480 seconds
1165866216 N * s0undt3ch_ s0undt3ch
1165866504 Q * matti Ping timeout: 480 seconds
1165866522 M * Bertl daniel_hozac: do you need a working version RSN (regarding the mapping/checks)?
1165866530 M * Bertl s/need/want/ :)
1165866537 M * daniel_hozac nah, CAP_MKNOD works fine for now ;)
1165866556 M * daniel_hozac i guess h01ger is rather interested though.
1165866571 M * daniel_hozac (for now) i only build my own packages.
1165866612 M * Bertl okay, I'll see what I can do ... I would like to make the checks lockless (i.e. use RCU)
1165866641 J * bronson ~bronson@66.160.177.208
1165866681 M * Bertl but I think I can do that later too, so I'll simply integrate the changes (slightly modified) and hand that back to you, yes?
1165866705 M * daniel_hozac sounds good to me.
1165866996 M * bronson Maybe a silly question...  Is there anything preventing me from booting a vserver from any directory on my filesystem?
1165867006 M * bronson Do I *have* to boot from /etc/vservers/.defaults/vdirbase?
1165867037 M * daniel_hozac no.
1165867047 M * bronson It's just the way the tools work today?
1165867083 M * daniel_hozac you can specify the directory on vserver ... build.
1165867106 M * bronson I mean, must all guests live in a single directory?
1165867112 M * daniel_hozac no.
1165867117 M * bronson Could I boot one from /var/vservers, another from /home, another from /tmp?
1165867124 M * daniel_hozac there's a reason the symlink is in the per-guest configuration.
1165867150 M * bronson Ah, I didn't see that.  Good deal.
1165867173 M * daniel_hozac you'll have to make sure to set the barrier on all the directories above the guest roots though.
1165867252 M * Bertl daniel_hozac: any 'obvious' reason why we cannot simply use S_IFCHR and S_IFBLK in our internal flags?
1165867269 M * daniel_hozac S_IFCHR isn't a bit value.
1165867269 M * daniel_hozac it's 6.
1165867273 M * Bertl i.e. just map in the S_IFMT?
1165867274 M * daniel_hozac a S_IFBLK is 4.
1165867284 M * Bertl yeah, but that should be fine, no?
1165867311 M * Bertl I mean, we cannot reasonably have entries matching both, no?
1165867361 M * Bertl (a single entry matching both that is)
1165867619 M * daniel_hozac hmm, no?
1165867633 M * daniel_hozac for defaults with no target, it makes sense, no?
1165867645 M * Bertl yes, but defaults are device less anyway
1165867670 M * Bertl i.e. they apply to both, or would you like to set independant
1165867683 M * Bertl defaults for char and blk? hmm, might make sense
1165867710 M * Bertl but in that case, we definitely need to pass the type to the kernel, which could otherwise be avoided
1165867712 M * daniel_hozac i figured we'd just want it as configurable as possible.
1165867722 M * daniel_hozac well, yeah.
1165867726 M * daniel_hozac but it could be optional, no?
1165867732 M * daniel_hozac like it is in my patch.
1165867765 J * matti matti@linux.gentoo.pl
1165867793 M * Bertl hmm, OTOH, we could simply deduce the 'type' from a given target, for target less cases, we only have flags which could be applied to both
1165867796 M * daniel_hozac seems like i forgot to set it for defaults and ACLs.
1165867845 M * daniel_hozac what about the target-less default?
1165867858 M * daniel_hozac i.e. no device, no target.
1165867869 M * Bertl would apply to both
1165867927 M * daniel_hozac yeah, i guess that's versatile enough.
1165867968 M * daniel_hozac i can't think of any reason why you'd want to give general access to character devices but not block devices, or vice versa..
1165868195 M * Bertl me neither, for now, I guess I'll add the internal support for separate entries, but avoid the external interface to that, and let the kernel decide, okay?
1165868252 M * daniel_hozac so mask out those values from the userspace supplied flags?
1165868281 M * Bertl I was more thinking of the following semantics:
1165868297 M * Bertl - get device and S_IFMT (from device)
1165868307 M * Bertl - get target and S_IFMT
1165868343 M * Bertl - if device and target and device_fmt != target_fmt error out 
1165868362 M * Bertl - if device and !target -> type = device_fmt
1165868373 M * Bertl - if target and !device -> type = target_fmt
1165868381 M * daniel_hozac right, that's what i had intended with my patch.
1165868395 M * Bertl - if !device and !target -> type = botj
1165868398 M * Bertl *both
1165868416 Q * mire Ping timeout: 480 seconds
1165868416 M * daniel_hozac (and i originally had that, but then i moved things around and that got removed...)
1165868434 M * Bertl okay, so that should suit our purpose, yes?
1165868442 M * daniel_hozac yeah, i think so.
1165868523 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch
1165869820 Q * sebastian 
1165869970 Q * michal` Ping timeout: 480 seconds
1165870083 Q * meandtheshell Quit: Leaving.
1165870605 J * Aiken ~james@tooax6-120.dialup.optusnet.com.au
1165870621 M * Bertl morning Aiken!
1165870641 M * Aiken hi
1165870711 J * michal` ~michal@www.rsbac.org
1165870779 M * Aiken no changes with my bug report but did at least seemed to find the patch that broke the kernel for me
1165870793 M * Bertl ah, great, what was it?
1165870829 M * Aiken the bug report http://bugzilla.kernel.org/show_bug.cgi?id=7662
1165870851 M * Aiken changes made to aoe between rc-2 & rc3
1165871022 Q * hardwire Quit: Coyote finally caught me
1165871050 J * hardwire ~hardwire@rdbck-7535.palmer.mtaonline.net
1165871151 M * hardwire adsf
1165871606 M * raw  http://linux-vserver.org/util-vserver:Howto_virtualize_an_exisiting_Linux_server <- why should sshd_config -> UsePAM be disabled on guest system? i realy want to know :)
1165871682 M * Bertl raw: have you tried without?
1165871693 M * Bertl i.e. without disabling it?
1165871716 M * Bertl as I mentioned earlier, I think that is just how the person writing the wiki page did it
1165871720 M * daniel_hozac raw: you'd have to ask the author.
1165871742 M * raw if i disable it it dont asks me for a password :/
1165871762 M * raw so i cant login
1165871771 Q * bonbons Quit: Leaving
1165871810 M * Bertl raw: well, I guess your setup doesn't have/use shadow passwords then
1165871827 M * Bertl otherwise, IIRC, sshd should use that for authentication
1165871856 M * raw okay :P
1165872663 Q * Medivh- Read error: Connection reset by peer
1165873394 Q * Piet Remote host closed the connection
1165873432 J * Piet hiddenserv@tor.noreply.org
1165873616 M * lomonaco Hi, I used the VServer Installation Fedora Core 5 Guide and everything is ok, but now I´m trying to compile the Asterisk without success
1165873672 M * lomonaco Does anyboy know how can i configure yum´s repository to include the kernel source. Thanks in advanced
1165873678 M * Bertl hello lomonaco! how so?
1165873793 M * lomonaco hi Bertl, thank you for your reply, I just wanna know how can I install the patched kernel to be able to compile the asterisk soruce
1165873801 M * lomonaco source, sorry
1165873828 M * Bertl ah, so you didn't install the kernel yet?
1165873848 M * daniel_hozac why did you use FC5? FC6 is current.
1165874004 M * lomonaco no Bertl, i download the following file kernel-2.6.18-1.2239.fc5.vs2.0.2.1.src.rpm but i got the following error
1165874019 M * daniel_hozac why do you need the source? kernel-devel should suffice.
1165874069 Q * Smutje Remote host closed the connection
1165874070 M * johnny nebuchadnezzar, did you setup the dummy interface in the host?
1165874090 M * lomonaco hi daniel_hozac, first thanks a lot for your excelent work and thanks too for sharing it to us
1165874103 M * Bertl johnny: ah, the dummy interface :)
1165874126 M * johnny well i've seen quite a few ways to handle the task i want to handle.. so i guess i'll use that
1165874142 M * johnny but what needs to be put in interfaces/1
1165874156 M * Bertl depends on what you want to do?
1165874157 M * johnny is it possible to start it in the vserver, and if so, should i?
1165874163 M * daniel_hozac no.
1165874186 M * johnny i just want something to act like 127.0.0.1
1165874192 M * daniel_hozac lomonaco: someone's got to do it :)
1165874193 M * johnny so i don't have to use my guest vserver ip
1165874196 M * daniel_hozac johnny: for binds, right?
1165874203 M * daniel_hozac i.e. a non-public address.
1165874205 M * johnny yes
1165874240 M * lomonaco [root@fedora asterisk]# cd zaptel-1.2.11; make
1165874241 M * lomonaco You do not appear to have the sources for the 2.6.18-1.2239.fc5.vs2.0.2.1 kernel installed.
1165874263 M * daniel_hozac lomonaco: have you installed the kernel-devel package?
1165874269 M * daniel_hozac lomonaco: that should work.
1165874300 M * johnny so daniel_hozac i should i proceed then? 
1165874309 M * johnny i setup dummy0 on the host at 192.168.10.1
1165874324 M * johnny so do i use the nodev file? and set that ip ?
1165874344 Q * cdrx Ping timeout: 480 seconds
1165874415 M * Bertl johnny: what is it used for?
1165874423 M * daniel_hozac yes, if you don't want to bring up the IP address with the guest.
1165874436 M * daniel_hozac i.e. you intend to add that to whatever configuration file your host uses to bring up addresses.
1165874438 M * johnny i want to bind postfix to it 
1165874449 M * johnny in my vserver
1165874457 M * johnny and then let assp talk to it
1165874461 M * lomonaco hi daniel, I´m trying right now
1165874462 M * johnny and be external 
1165874473 M * lomonaco just yum install kernel-dev ?
1165874477 M * Bertl johnny: be external?
1165874483 M * daniel_hozac lomonaco: right, yum install kernel-devel
1165874484 M * johnny assp will be external 
1165874490 M * johnny on port 25
1165874499 M * johnny postfix should never be exposed
1165874516 M * Bertl okay
1165874535 M * Bertl both are in the same guest?
1165874549 M * johnny yes
1165874557 M * Bertl then you can also add that ip to the guest
1165874566 M * Bertl i.e. let the tools create/add it
1165874577 M * lomonaco i´m using 2.6.18-1.2239.fc5.vs2.0.2.1 and the yum install kernel-devel brings me  2.6.18-1.2255.fc5.vs2.0.2.2.0.rc9.1
1165874578 M * johnny uhmm.. hmm
1165874584 M * lomonaco should I continue ?
1165874603 M * johnny so i load the dummy module and then set the net interface up where?
1165874607 M * johnny in the host or the vserver?
1165874609 M * daniel_hozac lomonaco: well, you should probably upgrade to that kernel too then.
1165874616 M * daniel_hozac lomonaco: i just pushed it earlier today.
1165874619 M * Bertl lomonaco: obviously your yum has more recent info 
1165874653 M * johnny Bertl, i remember you helped me ages ago with my vserver it seems
1165874658 M * johnny it's been up for over a year now
1165874664 M * Bertl great!
1165874683 M * Bertl johnny: just add the dummy0 and ip as interfaces/1
1165874712 M * johnny but not configure it in the host at all?
1165874722 M * Bertl should not be required
1165874739 M * Bertl when you bring the guest up, it will be initialized
1165874812 M * johnny is there a way to add it to a currently running vserver tho?
1165874848 M * daniel_hozac not if it only had one IP address before.
1165874854 M * Bertl johnny: probably not
1165874887 M * brcc bertl no one tried iptables yet
1165874888 M * brcc :(
1165874909 M * daniel_hozac brcc: really?
1165874913 M * Bertl brcc: hmm, you have to advertize it more
1165874924 M * johnny hmm.. guess i can try to restart it then..
1165874928 M * johnny that has been awhile :)
1165874933 M * brcc just one guy, sometime ago
1165874939 M * brcc i am almost quiting
1165874969 M * brcc on getting people to test it
1165874978 M * brcc gonna do it all myself, i need to do the iptables-restore and iptables-save
1165874980 M * brcc and we are done.
1165875168 M * Guy- brcc: what is it you are doing? iptables support for vservers?
1165875186 M * Guy- brcc: that is, a vserver could have its own firewall rules?
1165875294 M * brcc yes
1165875312 M * lomonaco thanks a lot, daniel_hozac and Bertl. I will start from the beggining using FC6. Can I use the FC5 Tutorial ?
1165875329 M * daniel_hozac lomonaco: yeah, just replace 5 with 6 where you see it :)
1165875331 M * brcc the experimental version already works
1165875445 M * daniel_hozac Medivh: RPMs should be ready within an hour, it's in the final stages now.
1165875467 J * meandtheshell ~markus@85-124-206-43.dynamic.xdsl-line.inode.at
1165875512 M * Bertl welcome meandtheshell!
1165875524 M * lomonaco thank you again Daniel, I can translate your tutorial to brazilian portuguese (Brazil), do you want ?
1165875533 M * meandtheshell Bertl: hi there
1165875580 M * Bertl lomonaco: the problem with translations usually is that they do not keep up with changes
1165875624 M * daniel_hozac lomonaco: what we'd appreciate even more would be a migration to the new wiki ;) (AFAIK noone has done it)
1165875725 Q * michal` Ping timeout: 480 seconds
1165875839 Q * DavidS Quit: Leaving.
1165875939 M * lomonaco no problems, Daniel, where can i find more information about how to do that ?
1165876056 Q * transacid helium.oftc.net osmosis.oftc.net
1165876056 Q * nebuchadnezzar helium.oftc.net osmosis.oftc.net
1165876056 Q * Curus helium.oftc.net osmosis.oftc.net
1165876056 Q * harry helium.oftc.net osmosis.oftc.net
1165876056 Q * eGnarF helium.oftc.net osmosis.oftc.net
1165876056 Q * derjohn helium.oftc.net osmosis.oftc.net
1165876056 Q * Loki|muh helium.oftc.net osmosis.oftc.net
1165876056 Q * mugwump helium.oftc.net osmosis.oftc.net
1165876056 Q * SNy helium.oftc.net osmosis.oftc.net
1165876056 Q * Medivh helium.oftc.net osmosis.oftc.net
1165876069 J * transacid ~transacid@transacid.de
1165876069 J * nebuchadnezzar ~nebu@zion.asgardr.info
1165876069 J * Curus ~Curus@kbhn-vbrg-sr0-vl209-213-185-8-10.perspektivbredband.net
1165876069 J * Medivh ck@paradise.by.the.dashboardlight.de
1165876069 J * harry ~harry@d54C2508C.access.telenet.be
1165876069 J * eGnarF ~bartek@bk.crystone.se
1165876069 J * derjohn ~derjohn@80.69.41.2
1165876069 J * Loki|muh loki@satanix.de
1165876069 J * mugwump ~samv@watts.utsl.gen.nz
1165876069 J * SNy d3f8ac7800@bmx-chemnitz.de
1165876088 Q * johnny charon.oftc.net cation.oftc.net
1165876088 Q * dna charon.oftc.net cation.oftc.net
1165876088 Q * shedi charon.oftc.net cation.oftc.net
1165876088 Q * Zaki_ charon.oftc.net cation.oftc.net
1165876088 Q * ||Cobra|| charon.oftc.net cation.oftc.net
1165876088 Q * duckx charon.oftc.net cation.oftc.net
1165876088 Q * yang charon.oftc.net cation.oftc.net
1165876088 Q * Osgiliath charon.oftc.net cation.oftc.net
1165876088 Q * TheSeer charon.oftc.net cation.oftc.net
1165876088 Q * Vudumen charon.oftc.net cation.oftc.net
1165876088 Q * renihs charon.oftc.net cation.oftc.net
1165876088 Q * Rich_Estill charon.oftc.net cation.oftc.net
1165876090 J * johnny ~johnny@ip70-160-117-211.hr.hr.cox.net
1165876090 J * dna ~naucki@202-205-dsl.kielnet.net
1165876090 J * shedi ~siggi@inferno.lhi.is
1165876090 J * Zaki_ ~Zaki@88.213.58.243
1165876090 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl
1165876090 J * duckx ~Duck@tox.dyndns.org
1165876090 J * yang ~yang@yang.sponsor.oftc.net
1165876090 J * Osgiliath ~osgiliath@vdebian.org
1165876090 J * TheSeer ~theseer@border.office.salesemotion.net
1165876090 J * Vudumen c28212698f@perverz.hu
1165876090 J * renihs ~penguin@83-65-34-34.arsenal.xdsl-line.inode.at
1165876090 J * Rich_Estill ~restill@c-24-11-195-139.hsd1.mi.comcast.net
1165876094 M * Guy- brcc: sounds good; I'm vaguely interested, especially because I so far thought this wasn't going to be done, ever
1165876118 M * daniel_hozac lomonaco: http://wiki.linux-vserver.org/Wiki_Team i guess.
1165876119 M * Guy- brcc: is it in the wiki somewhere? if not, it should be, and I think people would try it alright
1165876132 M * daniel_hozac Guy-: it was on the ML.
1165876158 M * daniel_hozac with a link to http://www.virtuaserver.com.br/forum/viewtopic.php?p=215
1165876280 J * michal` ~michal@www.rsbac.org
1165876304 M * brcc grerat
1165876317 M * brcc check the forum, i gotta update some fixes i have done
1165876322 M * brcc but that version already works
1165876381 Q * hardwire helium.oftc.net strange.oftc.net
1165876381 Q * Aiken helium.oftc.net strange.oftc.net
1165876381 Q * matti helium.oftc.net strange.oftc.net
1165876384 J * hardwire ~hardwire@rdbck-7535.palmer.mtaonline.net
1165876384 J * Aiken ~james@tooax6-120.dialup.optusnet.com.au
1165876384 J * matti matti@linux.gentoo.pl
1165876616 M * johnny this is fun.. now my vserver doesn't start up .. been such a long time.. :(
1165876626 M * daniel_hozac hehe.
1165876632 M * daniel_hozac why not?
1165876697 M * Guy- daniel_hozac: thanks
1165876773 M * johnny make: *** [.twain.localmomentum.net.stamp] Error 255
1165876773 M * johnny make: Target `all' not remade because of errors.                          [ !! ]
1165876854 J * anonc ~anonc@staffnet.internode.com.au
1165876946 M * daniel_hozac johnny: and before that?
1165877345 M * johnny well i started it with my distro's init scripts
1165877350 M * johnny so i just retried
1165877361 M * johnny now i'm getting this one
1165877362 M * johnny chbind: kernel does not provide network virtualization
1165877401 M * Bertl could be that you updated the tools at some point?
1165877409 M * johnny yeah
1165877431 M * Bertl and the kernel?
1165877440 M * daniel_hozac you have the legacy version enabled in the kernel, and no legacy support in the utils.
1165877441 M * johnny uhmm no
1165877447 M * johnny yeah i just noticed that
1165877454 M * johnny guess the old package
1165877460 M * johnny had it enabled
1165877474 M * johnny this is the earliest version available in the tree
1165877475 M * johnny 0.30.210
1165877497 M * johnny i'm using 2.0.1
1165877503 M * johnny for the kernel patch
1165877524 M * johnny i really would like to reboot this machine.. and install a newer kernel
1165877536 M * johnny but i really hate remote reboots without a console :(
1165877574 M * johnny i guess i read enable legacy as also enable legacy
1165877627 M * johnny not only enable legacy, otherwise i wouldn't have chosen it
1165877645 M * daniel_hozac what?
1165877671 M * johnny like.. enable legacy and the normal api
1165877675 M * johnny at the same time
1165877680 M * johnny not 1 vs other
1165877698 M * daniel_hozac yes, that's what it does.
1165877704 M * johnny which one?
1165877708 M * daniel_hozac only you enabled the legacy version.
1165877713 M * daniel_hozac which makes the kernel appear older.
1165877717 M * johnny if it's 1 vs other.. then the new version 
1165877724 M * johnny should have an option too
1165877732 M * daniel_hozac the new version is always enabled.
1165877732 M * johnny hmm.. well that sucks..
1165877735 M * johnny oh ok
1165877739 M * daniel_hozac yeah, so why did you pick it?
1165877749 M * johnny cuz i assumed it meant both old and new would work
1165877757 M * johnny since thee's no indication that it meant that in the kernel config
1165877757 M * daniel_hozac not the legacy version.
1165877762 M * daniel_hozac yeah there is.
1165877770 M * daniel_hozac the help message is _very_ explicit.
1165877774 M * johnny yeah in help ..
1165877778 M * johnny perhaps
1165877808 M * daniel_hozac +         This will probably disable some features of newer tools
1165877808 M * daniel_hozac +         so better avoid it, unless you really, really need it
1165877808 M * daniel_hozac +         for backwards compatibility.
1165877813 P * stefani I'm Parting (the water)
1165877825 M * johnny hmm.. guess i should try to restart my host
1165877890 M * Bertl hmm, why not turn legacy on?
1165877915 M * johnny i can.. but i'm going to have to make sure this thing reboots eventually :)
1165877934 M * Bertl nah, in the tools, without rebooting
1165877939 M * johnny yeah i know that 
1165877950 M * johnny but.. i was just saying that i do have to reboot the host eventually.. just to make sure it still does
1165877958 M * johnny i've upgraded so much since then
1165877965 M * Bertl ah, yes, but with a console then, no?
1165877978 M * johnny it's in some box in tampa.. bout 700 miles away
1165877988 M * johnny so no
1165877998 M * Bertl means just that you need a longer cable :)
1165878022 M * johnny lol
1165878029 M * johnny this box was originally redhat9
1165878042 M * johnny now it's a gentoo install
1165878042 M * hardwire oh lord!
1165878051 M * hardwire a friend of mine is too  lazy to recompile his own kernel
1165878061 M * johnny i recompile kernels all the time..
1165878063 M * daniel_hozac who isn't? :)
1165878065 M * hardwire so I decided to do a test and see if edgy can even do this with its own packages
1165878066 M * johnny but it's scary to do remote..
1165878067 M * Bertl hardwire: happens all the time ...
1165878069 M * hardwire no
1165878076 M * hardwire Bertl: edgy is so broken for vserver
1165878082 M * daniel_hozac yes, it is.
1165878088 M * daniel_hozac Ubuntu in general needs a maintainer.
1165878088 M * hardwire you can't patch edgy kernel source from the vserver site
1165878098 M * hardwire you can't use kernel-patch-vserver
1165878098 M * hardwire fine
1165878099 M * johnny os what's the option i need to pass to --configure ?
1165878101 M * hardwire I will fucking do it
1165878102 M * hardwire this is stupid
1165878111 M * daniel_hozac johnny: --enable-apis=NOLEGACY
1165878112 Q * lomonaco Quit: Download IceChat at www.icechat.net
1165878124 M * johnny NOLEGACY?
1165878124 M * Bertl johnny: don't think about the option :)
1165878151 M * daniel_hozac it enables all but the really old API.
1165878158 M * daniel_hozac (which is called legacy)
1165878163 M * johnny this is what i see in vserver-stat 
1165878164 M * johnny  Available APIs: v13,net
1165878174 M * Bertl which is not what we call legacy now
1165878188 M * johnny yeah i knw it's not right :0
1165878198 M * johnny i just want to choose the right thing now :)
1165878205 M * Bertl v13,net is the most recent API
1165878224 M * Bertl you need the one older one, which is what you will get with NOLEGACY
1165878232 M * johnny hmm ok
1165878235 M * daniel_hozac well, v13,net,v21 is the most recent API :)
1165878246 M * Bertl hmm, right :)
1165878256 M * johnny i guess it doesn't make sense from an outsider perspect to have legacy enabled by NOLEGACY :)
1165878282 M * hardwire great.. now he is going to write off vservers for good since he can't get it working as a test on his workstation
1165878283 M * hardwire grr
1165878283 M * daniel_hozac what the kernel calls legacy is called compat in the utils.
1165878376 M * johnny gonna go pick up the gf from work.. i'll brb
1165878870 M * Bertl hardwire: well, every man is the architect of his own fortune ...
1165878946 M * hardwire I think I am gonna come up with a better kernel-patch-vserver and call it good for now
1165878961 M * hardwire and then start releasing linux-image-...-vserver packages
1165878970 M * hardwire and see if I can become an official package manager
1165878971 M * daniel_hozac like Debian is already doing?
1165878980 M * hardwire dude
1165878985 M * hardwire exactly like debian is doing
1165878993 M * hardwire but ubuntu ignores debian kernelage
1165879000 M * daniel_hozac but i think that would be good.
1165879005 M * hardwire I ponder if I can just install a debian kernel package
1165879007 M * Bertl hardwire: for ubuntu?
1165879010 M * daniel_hozac Ubuntu is in desperate need of a vserver maintainer.
1165879012 M * hardwire Bertl: yeh
1165879018 M * hardwire I think I may put on a red cape for the cause
1165879021 M * daniel_hozac (as nobody seems to really care for it)
1165879022 M * hardwire cause it bugs the shit out of me
1165879028 M * Bertl hardwire: well, we all would appreciate an unbuntu maintainer ...
1165879040 M * hardwire its not really that high of demand afaik
1165879047 M * hardwire but it does cause a lot of bitchin
1165879065 M * hardwire for instance.. why did somebody start with xen and stop there
1165879164 M * daniel_hozac Bertl: any progress on the patch? :)
1165879206 M * hardwire whats the problem with enabling the vserver cpu prioritization by default?
1165879211 M * hardwire it seems the debian kernels left that out
1165879222 Q * dna Quit: Verlassend
1165879235 M * daniel_hozac i think the rationale was "it's disabled by default upstream".
1165879263 M * hardwire why is it disabled by default upstream?
1165879274 M * hardwire what issues pop up :)
1165879282 M * daniel_hozac because it adds a slight scheduler overhead.
1165879287 M * hardwire ohnoes
1165879318 M * hardwire poor babys!
1165879447 M * hardwire somebody went overboard with xen and vserver patches combined for the debian packages
1165879453 M * hardwire I think a little crazy actually
1165879458 M * daniel_hozac why's that?
1165879464 M * hardwire its nice.
1165879467 M * hardwire you can test both
1165879478 M * hardwire I haven't run into many people running both however
1165879487 M * hardwire or a server that doesn't mind throwing its resources around like that
1165879497 M * hardwire a xen domU kernel with vserver makes sense
1165879511 M * hardwire but the why not factor wins
1165879511 M * daniel_hozac ... that's what it is.
1165879520 M * hardwire daniel_hozac: thats for dom0
1165879536 M * hardwire vserver and xen patches for the host machine
1165879539 M * daniel_hozac i doubt it.
1165879551 M * hardwire then you doubt the dark side.
1165879555 A * hardwire raises hands
1165879572 M * daniel_hozac i'm quite certain it's the domU kernel that has vserver.
1165879573 M * daniel_hozac waldi?
1165879699 M * hardwire ooh its tiny!
1165879705 M * hardwire I think you are on to something
1165879721 Q * yarihm Quit: Leaving
1165879776 M * Bertl daniel_hozac: how does this look for you? http://vserver.13thfloor.at/Experimental/delta-dmap-feat04.diff
1165879776 M * hardwire yeh its a kernel
1165879782 M * hardwire for the domU
1165879784 M * hardwire nuhr
1165879793 M * hardwire that makes a weeny bit more sense now
1165879817 M * daniel_hozac Bertl: no DATTR_REMAP?
1165879860 M * daniel_hozac ah, nevermind, that's implicit from the function name.
1165879925 M * Bertl yes, we can make it explicit once we combine the functions
1165879937 M * Bertl I wanted to have them separate for now, for debugging
1165879952 M * Bertl (and maybe they should stay separate)
1165879981 M * Bertl but you are right, it should check for the REMAP in the lookup
1165880054 M * daniel_hozac what's the second hunk for __lookup_mapping for?
1165880062 M * daniel_hozac the if (target) ... stuff.
1165880092 M * Bertl that is the current default
1165880106 M * Bertl will be replaced by a 'real' default entry soon
1165880116 M * daniel_hozac okay, that's what i figured.
1165880273 M * hardwire the debian kernel worked fine
1165880285 M * hardwire afaik nothing is sacrificing sheep in the background
1165880304 M * daniel_hozac i guess the "mode 0 means any" isn't implemented yet, right?
1165880567 Q * lilalinux Remote host closed the connection
1165880718 M * Bertl daniel_hozac: nope
1165880752 M * Bertl but it is only relevant for the defaults, which are not implemented either
1165880759 M * daniel_hozac yeah.
1165880804 M * Bertl we probably should start tagging 'special' inodes with xid
1165880829 M * Bertl and when they are accessed from a different context, purge and remap them
1165880853 M * daniel_hozac yeah, that'd probably make testing it easier too.
1165880855 M * Bertl (or just block the access)
1165880966 M * Bertl for the testing, I somehow remember there was a way to flush the inode cache, but I cannot remember how
1165881499 M * daniel_hozac btw, looks like you missed the __user_device's in the CONFIG_COMPAT case.
1165881525 M * daniel_hozac and the mode variables.
1165881531 M * Bertl indeed
1165881576 Q * DreamerC Remote host closed the connection