WebHosting Paid by #1Payday.Loans
[02:17] esden (i=esdentem@pallas.crash-override.net) left irc: Remote closed the connection
[02:57] Freak (n=freak@erleuchtet.org) left irc: Read error: 60 (Operation timed out)
[02:59] Freak (n=nfreak@erleuchtet.org) joined #rocklinux.
[03:48] netrunner (n=andreas@anvame.net) left irc: Read error: 110 (Connection timed out)
[04:04] netrunner (n=andreas@anvame.net) joined #rocklinux.
[08:09] <blindcoder> moin
[09:00] <blindcoder> wow, 193 patches...
[09:22] blindcoder (n=blindcod@dslb-084-059-044-045.pools.arcor-ip.net) left irc: "brb"
[09:47] blindcoder (n=blindcod@tor/session/x-7f98edcbeec9a5b1) joined #rocklinux.
[09:50] netrunner (n=andreas@anvame.net) left irc: "Reconnecting"
[09:51] netrunner (n=andreas@anvame.net) joined #rocklinux.
[09:51] <netrunner> blindcoder: I changed dmmount.sh so it forcedly tries to mount ext3 first. otherwise mount always uses ext2 for my ext3 fs
[09:51] <blindcoder> netrunner: nice. can you send me the new version?
[09:52] <netrunner> blindcoder: it is easyer to tell you :)
[09:52] <netrunner> blindcoder: just do mount -t ext3 ..... || \ mount .....
[09:52] <netrunner> oh, the backslash just masks a newline here.
[09:53] <blindcoder> ah, okay
[09:54] <blindcoder> if ! ( mount -t ext3 /dev/mapper/${encryptedname} ${mountpoint} || mount -t ext3 /dev/mapper/${encryptedname} ${mountpoint} ) ; then
[09:54] <blindcoder> that should work, no?
[09:54] <blindcoder> ehm
[09:54] <blindcoder> without the second -t ext3 of course
[09:55] <netrunner> why so complicated? :)
[09:56] <blindcoder> netrunner: I've got a lot of local changes to dmscripts here
[09:56] <blindcoder> netrunner: error checking, mostly
[09:56] <blindcoder> which I'm about to upload
[10:42] <blindcoder> netrunner: dmscripts update done
[10:49] <blindcoder> urgs
[10:49] <blindcoder> proxychains needs a 1.7 MB patch to build...
[11:27] esden (i=esdentem@pallas.crash-override.net) joined #rocklinux.
[11:30] <daja77> so dsl works now
[11:39] <blindcoder> nice :)
[11:41] Action: daja77 agrees
[11:43] <netrunner> https://www.maxxx-flash.de/videos/lights.wmv
[11:48] <daja77> hehe
[12:03] blindcod1r (n=blindcod@tor/session/x-a0afd13704bab025) joined #rocklinux.
[12:04] blindcoder (n=blindcod@tor/session/x-7f98edcbeec9a5b1) left irc: Nick collision from services.
[12:04] Nick change: blindcod1r -> blindcoder
[12:33] [raphael] (n=raphael@lehrenetz.fh-wels.at) joined #rocklinux.
[12:59] <blindcoder> blindcoder@ceres:~$ scp scavenger.homeip.net:/THINGS_TO_CHECK blindcoder@pallas.crash-override.net:/foo/ba\:l/baz
[12:59] <blindcoder> Resolving scavenger.homeip.net through tor... 84.59.36.64
[12:59] <blindcoder> Resolving pallas.crash-override.net through tor... 213.239.220.170
[12:59] <blindcoder> scp "84.59.36.64:/THINGS_TO_CHECK" "blindcoder@213.239.220.170:/foo/ba:l/baz"
[12:59] <blindcoder> MUAHAHAHA
[13:04] <th> what ssh version is this?
[13:04] <th> that was not possible before
[13:04] <th> remote 2 remote copy...
[13:06] <blindcoder> I don't think it's possible now
[13:06] <blindcoder> it's just a test case
[13:06] <blindcoder> even though the manpage suggests it works
[13:07] <th> really?
[13:07] <blindcoder> from the manpage
[13:08] <blindcoder> scp [-1246BCpqrv] ... [[user@]host1:]file1 [...] [[user@]host2:]file2
[13:08] <blindcoder> Any file name may contain a host and user specification to indicate that
[13:08] <blindcoder> the file is to be copied to/from that host. Copies between two remote
[13:08] <blindcoder> hosts are permitted.
[13:13] <blindcoder> just tested, it works
[13:14] <blindcoder> openssh 3.9p1 0
[13:19] [anders-mcafee] (n=anders@213.208.70.155) left irc: Read error: 104 (Connection reset by peer)
[13:35] <blindcoder> okay, I have now covered: irssi, bittorrent, ssh, scp and everything that uses HTTP_PROXY and http_proxy
[13:39] youngcoder (n=codemast@cpe-24-160-27-85.sw.res.rr.com) joined #rocklinux.
[13:43] madtux (i=miguel@pf0.hostarica.com) joined #rocklinux.
[13:43] <madtux> hello.
[13:43] <daja77> hi miguel
[13:43] <blindcoder> moin madtux
[13:44] <madtux> Blindy :)
[13:44] <blindcoder> madtux: no, I'm just a random guy hiding behind tor :P
[13:46] <madtux> :P
[13:51] <blindcoder> okay, telnet also covered
[13:52] <madtux> lets see what isn't
[13:52] <blindcoder> heh, a LOT
[13:53] <blindcoder> I'm writing a source file (. tor_wrapper) so that various applications are transparently routed through the tor network
[13:54] <blindcoder> so far I have: telnet, ssh, scp, irssi, bittorrent and everything that uses HTTP_PROXY and http_proxy
[13:54] <madtux> cool.
[13:54] <blindcoder> all you gotta do is just source that file from your ~/.bashrc
[13:54] <blindcoder> and then just call "irssi", "ssh" or whatever
[13:57] <blindcoder> with the recent idiocy here in europe, you really gotta start doing something
[14:00] <netrunner> While reading GEM file tor: No such file or directory
[14:00] <clifford> what is your current directory?
[14:01] <netrunner> who's? mine? in which of my many open windows? :)
[14:03] Action: daja77 likes to push the gnome update now, even when it s only partly tested to integrate the tons of new updates in a new build
[14:04] <blindcoder> hmm
[14:04] <blindcoder> does safe_useradd create the $HOME directory?
[14:06] <netrunner> no
[14:08] <blindcoder> damn :(
[14:08] <netrunner> do you see any directory in build/foo/home? :)
[14:08] <blindcoder> ehm... no
[14:09] <blindcoder> okayokay, my bad :P
[14:09] <blindcoder> the problem is that the tor service needs a $HOME
[14:09] <blindcoder> and it needs to be owned by himself
[14:09] <netrunner> /var/<service>?
[14:10] <clifford> blindcoder: services should have their home in something like /var/lib/foobar
[14:10] <clifford> .. but never in /home
[14:10] <blindcoder> clifford: yeah, that's where tor has its $HOME
[14:10] <blindcoder> /var/lib/tor
[14:10] <clifford> netrunner: /var/<service> is forbidden according to FHS
[14:10] <blindcoder> at least, that's where I put it
[14:10] <daja77> clifford: is it ok to push that gnome update and doin some fixing later on?
[14:10] <netrunner> clifford: sorry, was a quick guess :)
[14:11] <clifford> daja77: how much fixing do you expect?
[14:11] <netrunner> daja77: sm send it, give it a note that it is partially untested. then I can smap it and run a build :)
[14:11] <netrunner> clifford: btw: I still cannot build 1-glibc23
[14:12] <daja77> clifford: I already did some, but after the latest patches my tree is massively outdated and I wanted to sync again
[14:12] <netrunner> clifford: I get some assembler 'already defined' errors
[14:12] <daja77> the old gnome was broken for gcc4 anyway ...
[14:12] <clifford> netrunner: this is not sufficient information for helping you with glibc..
[14:13] <clifford> daja77: I will start a build with >150 pathes applied today.
[14:13] <clifford> .. that will pretty sure have already enough regressions.
[14:13] <netrunner> clifford: https://andreas.anvame.net/content/download/1-glibc23.err
[14:13] <blindcoder> so we probably won't have a recent Crystal handout for 22C3 :P
[14:13] <daja77> clifford: ok so you can leave out that patch, but netrunner can do some testing
[14:14] <netrunner> daja77: depends on if I will get past 1-glibc23 next time :)
[14:15] <daja77> ^^
[14:16] <clifford> netrunner: that one is new to me. I'm pretty sure it has nothing to to with NPTL
[14:16] <netrunner> hm. maybe I should commit local changes and do sm recreate to be in sync with you again ...
[14:17] <clifford> I did a generix x86 build - you did a pentium3 build. maybe that is the difference..
[14:17] <netrunner> clifford: I see that was with gcc 3.4.4 ... have you tested this compiler?
[14:17] <daja77> google tells me this is a binutils related bug
[14:17] <netrunner> binutils-2.16.90.0.3
[14:17] <daja77> https://sourceware.org/ml/binutils/2005-11/msg00208.html
[14:20] <daja77> damn why the hell that acl crap got into that patch
[14:22] <clifford> "sm send"-ing the last package update chunk.
[14:22] <daja77> hm?
[14:24] <netrunner> daja77: what acl crap got in which pach?
[14:25] <daja77> uncommitted changes on acl.conf
[14:25] <daja77> sm synching
[14:29] <blindcoder> https://shellscripts.org/project/toraliases
[14:29] <daja77> nice
[14:31] <blindcoder> a fscking necessity by now
[14:34] <clifford> wow.
[14:34] <clifford> 176 patches in this journal
[14:45] <demian> hi
[14:46] Action: netrunner sm synced again
[14:46] <netrunner> have I posted this already? https://5ka.mipt.ru/~serge/qwe.jpg
[14:46] <clifford> SMP: ping!
[14:49] <daja77> *gg* netrunner
[14:51] madtux (i=miguel@pf0.hostarica.com) left irc: "Leaving"
[14:52] markuman (n=supermar@p509257D8.dip.t-dialin.net) joined #rocklinux.
[15:11] <SMP> clifford: pong
[15:15] <clifford> SMP: give me a sec..
[16:07] <clifford> SMP: now I'm free again..
[16:07] <clifford> still around?
[16:08] <blindcoder> that was a looooong second ;)
[16:08] <SMP> clifford: yup
[16:08] <clifford> regarding glibc:
[16:08] <clifford> did you find anything out about the ld-linux.so thing?
[16:13] <SMP> clifford: partly. I now know that the ld.so.cache can list the "same" library multiple times, each differenciated by their embedded hwcap (bit field) and declared OS ABI level. both seem to be fields in the ELF header.
[16:14] <SMP> clifford: the OS ABI comes from, e.g. --enable-kernel= at glibc configure
[16:15] <SMP> clifford: it influences the availability of certain syscalls as well as the if the vsyscalls mechanism or straight int $80 is chosen
[16:15] <clifford> but how is this ld-linux.so feature activated? I've straced my ld-linux.so and it doesn't try to find stuff is the tls, etc subdirs.
[16:16] <clifford> .. and ldconfig isn't either.
[16:16] <SMP> yeah, ld-linux only consults the cache
[16:16] <SMP> I'm trying to find hints in the ldconfig source ..
[16:17] <SMP> could be the subdirs need to be explictly given to ldconfig, either compile-time or in ld.so.conf
[16:17] <clifford> ad ld.so.conf: no. at least debian doesn't have such entries there.
[16:18] <SMP> yeah, I know. so probably compiled in
[16:18] <SMP> I'll strace ldconfig right away. sec.
[16:18] <clifford> I did already. Mine doesn't looks for those subdirs.
[16:19] <clifford> btw: what libs (libdir) should be built with what options? (on x86)
[16:20] <SMP> but I'm trying on a system that uses such libs ;)
[16:20] <clifford> ah!
[16:22] <clifford> SMP: id ld-linux.so that does the magic
[16:22] <clifford> (tested on debian)
[16:23] <SMP> no, can't confirm
[16:24] <SMP> this here ldconfig looks in tls and i686 subdirs of every standard and extra dir it has been configured to use
[16:25] <SMP> if you're seeing ld-linux.so doing that, it's probably because you have LD_LIBRARY_PATH set
[16:25] <clifford> SMP: www.clifford.at/priv/debian-ld-linux-strace.txt
[16:25] [raphael] (n=raphael@lehrenetz.fh-wels.at) left irc: "using sirc version 2.211+KSIRC/1.3.12"
[16:25] <clifford> (no LD_* in my environment)
[16:25] <SMP> ah, ok
[16:26] <SMP> yet, if you set LD_LIBRARY_PATH you will see ld-linux.so look for those subdirs for every path in LD_LIBARY_PATH
[16:27] <SMP> your ld-linux.so probably does that because the cache lists /lib/tls/librt.so.1, which doesn't exist, so ld-linux goes looking for it
[16:28] <clifford> and it looks like ldconfig is indexing _everything_ it finds and adds the "hwcap" bitmaps to the ld.so.cache entries.
[16:28] <clifford> .. at least here.
[16:28] <SMP> what do you mean everything?
[16:29] <clifford> it also includes stuff in ls.so.cache which would not work with the current hwcap settings.
[16:29] <SMP> yes, this is the expected behaviour
[16:29] <clifford> (which in fact makes perfectly sense if e.g. various machines are sharing one root directory)
[16:30] <SMP> it's ld-linux that makes the decision at runtime
[16:30] <SMP> note $LD_ASSUME_KERNEL ...
[16:33] <SMP> clifford: are we d'accord about what we said so far?
[16:34] <clifford> yup.
[16:34] <clifford> expect for $LD_ASSUME_KERNEL - its not in my ld.so man page.. ;-)
[16:36] <clifford> netrunner: https://www.rocklinux.net/submaster/smadm.cgi?i=2005121516363704311
[16:37] <SMP> clifford: so what we should do (first) is IMHO: foreach kernel-2.4.0-or-so-ABI-no-optimize-LinuxThreads $more_ABI_and_cpu_optimization ; do build_and_install_a_glibc ; done
[16:38] <clifford> yes.
[16:38] <SMP> question would be, what's in $more_ABI_and_cpu_optimization.
[16:38] <SMP> your strace lists many interesting combinations ;-)
[16:38] <clifford> question is if we want to make it configurable in general
[16:39] <clifford> everything what is embedded would only need one glibc variant
[16:40] <SMP> agreed. so we need a $ROCKCFG_GLIBC_DEFAULT_ABI_AND_OPT and $ROCKCFG_ABI_AND_OPT_LIST
[16:40] <SMP> insert _GLIBC_ in the latter
[16:40] <clifford> bo-eh! the glibc debian diff is >1MB ..
[16:41] <clifford> yep.
[16:41] <blindcoder> clifford: proxychains is 1.8 MB :P
[16:41] <SMP> 1MB? you never saw the RedHat diff. it is (was) larger than the original compressed glibc sources ;->
[16:42] <SMP> (granted, most of which was just fucked up ChangeLog renaming)
[16:43] <SMP> clifford: then we make a list of ABI/CPU that we support, throw out everything that doesn't apply to the current $ROCKCFG_ARCH and let the user have the final choice
[16:44] <clifford> what is this "abilist/" directory in the glibc source tree?
[16:45] <SMP> I don't know, but it looks interesting
[16:49] <clifford> are linuxthreads and nptl abi compatible??
[16:50] <SMP> as far as I understand yes
[16:51] <clifford> ok. found it: www.clifford.at/priv/debian-glibc-i386.mk
[16:52] <clifford> I'd suggest that the non-default-fallback libcs are package forks.
[16:52] <clifford> e.g.: glibc23 glibc23-i686 glibc23-tls etc.
[16:52] <clifford> that makes it possible to install them seperately and also build them seperately
[16:53] <SMP> sounds reasonable
[16:54] Action: clifford is afk for a few minutes.
[17:11] <clifford> back.
[17:12] <clifford> I think I know how to make it..
[17:12] <SMP> hm?
[17:12] <clifford> we add two text strings to glibc23 config (as you suggested)
[17:12] <clifford> I'd prefer the wording "flavors" because that doesn't limit the meaning of it.
[17:13] <clifford> one for the default libc and one for the additional ones (whitespace spererated)
[17:13] <clifford> this does the package forking with -<flavor_name> added to tha package name
[17:13] <clifford> ..
[17:14] <clifford> the headers are always the same, right?
[17:14] <clifford> .. so we only need to install *.so files for the additional libcs.
[17:15] <SMP> yes, except we should have a look at the pthreads headers
[17:15] <clifford> .. hmm.
[17:15] <SMP> they do differ between LinuxThread and NPTL
[17:15] <SMP> but we only need the NPTL version
[17:16] <clifford> but the nptl libc shouldn't be the default libc.
[17:16] <clifford> that makes it a bit more complicated.
[17:18] <SMP> that's something we we'll have to look into once we can build other flavours and see what comes out
[17:19] <clifford> where is the source for the dynamic loader ???
[17:19] <SMP> another thing: I'd say we need to be a little smart and take ROCKCFG*OPT into account
[17:20] <clifford> elf/dl-* ?
[17:20] <SMP> dunno
[17:21] <daja77> clifford: iirc yes
[17:21] <SMP> if we optimize for i686 by default, then we don't need an i468 libc in /lib plus another one for i686 in /lib/i686
[17:22] <SMP> on the other hand, there may be circumstances when you want a nontls libc even when you're on i686
[17:22] <SMP> Xen comes to mind ...
[17:22] <clifford> yup. then you set the string for the default flavor to "i686" and let the one for the addition flavors empty.
[17:23] <clifford> the default values for the flavors would be architecture dependent.
[17:26] <SMP> no, I don't think i686 opt alone should imply NPTL. so even with i686 as default opt. you would have at least i686-tls as addnl. flavour
[17:27] <clifford> but if you only plan to run 2.6 kernels ..
[17:27] <SMP> then you can still find yourself on a Xen system that slows to a crawl when you use glibc's TLS layout ;)
[17:29] <SMP> unless you happen to have a CPU that does Vanderpool or Pacifica ;-)
[17:46] [raphael] (n=raphael@host-84-235-196-162.satlynx.net) joined #rocklinux.
[17:59] <clifford> hmm.. other distributions definitely don't have easier to understand glibc build scripts than we have.
[18:10] <SMP> plus, the .spec file language can only jump forward and has no procedures
[18:17] <clifford> argwl! texinfo brainwashes the environment using "env -i"
[18:18] <clifford> VERY bad for the install wrapper..
[18:42] <clifford> the massive-updates test build now reached stage 2.
[18:43] <clifford> 3 errors in 37 package builds so far (glibc, texinfo and bison)
[18:56] <demian> umm.. what does a SEARCH http command means?
[18:56] <demian> 201.239.116.197 - - [14/Dec/2005:20:00:44 -0600] "SEARCH /\x90\x02\xb1\x0...
[18:57] <clifford> iirc thats a WebDAV method
[18:59] <demian> well.. the data suplied to it was too long that apache said: request failed: URI too long (longer than 8190)
[18:59] <SMP> demian: old IIS exploit
[18:59] Action: SMP off
[19:00] <demian> ok,
[19:37] <netrunner> argl. someone sends spam to russia with my adress :)
[19:37] <netrunner> tomorrow they will kick me out :)
[19:43] <daja77> clifford: ack the same here
[19:44] <daja77> so I guess it is not gcc4 related
[20:06] <clifford> daja77: https://www.rocklinux.net/submaster/smadm.cgi?i=2005121520020325183
[20:06] <clifford> .. this is not needed for gcc4 ?
[20:07] <clifford> the build here is in stage 3 now. So I think I have left the greates difficulties behind.
[20:08] <clifford> (debugging gcc without an editor and with a cross-built readline-less bash is not funny)
[20:08] <daja77> clifford: i just applied these patches mom
[20:18] <daja77> clifford: texinfo and bison now built with gcc4
[20:24] Action: clifford is going home now. cu tomorrow.
[20:24] <daja77> cu
[21:08] markuman (n=supermar@p509257D8.dip.t-dialin.net) left irc: "https://www.linwiki.org - wiki around os linux!! #linwiki"
[21:34] menomc (n=amery@200.75.27.109) joined #rocklinux.
[21:35] mnemoc (n=amery@200.75.27.40) left irc: Nick collision from services.
[21:35] Nick change: menomc -> mnemoc
[22:11] <blindcoder> re
[22:14] <netrunner> wow, finally after 3 month we have 100MBit
[22:15] <blindcoder> ass :P
[22:15] <blindcoder> netrunner: open a tor server :D
[22:15] Action: mnemoc experience a cool 64kbps link
[22:19] <netrunner> well, the internet shouldn't be that fast, but at least local network.
[22:20] SteffenP (i=steffen@p54996045.dip.t-dialin.net) joined #rocklinux.
[23:15] kasc (n=kasc@dslb-084-060-124-217.pools.arcor-ip.net) left irc: Read error: 104 (Connection reset by peer)
[23:24] kasc (n=kasc@dslb-084-060-102-247.pools.arcor-ip.net) joined #rocklinux.
[23:27] SteffenP (i=steffen@p54996045.dip.t-dialin.net) left irc: "https://www.bomberclone.de"
[00:00] --- Fri Dec 16 2005