irclog2html for uclibc on 2002.08.19

06:11:19anderseeBZFlag: ibot is mostly useful for insulting people.. ;)
06:11:24andersee: what?
06:11:24anderseeibot: welcome
06:11:29andersee is nothing but an onion-eyed petrification of pribbling dog vomit.
06:11:29BZFlagibot insult andersee
06:11:32andersee: what?
06:11:32anderseeibot: your heard me
06:11:44andersee: what?
06:11:44anderseeibot: evil green frogs!
06:11:49rumour has it uclibc is at uclibc.org
06:11:49BZFlagibot uclibc
06:11:56okay, BZFlag.
06:11:56BZFlagibot uclibc is also #uclibc
06:14:30aaronlbzflag! hello
06:14:52aaronli think i noticed the other day that our PGP keys are two hops appart
06:15:05andersee is nothing but an antique coagulation of boil-brained pods.
06:15:05aaronlibot: insult andersee
06:15:11Since Mon Aug 19 07:10:29 2002, there have been 2 modifications and 3 questions.  I have been awake for 4 minutes, 42 seconds this session, and currently reference 84016 factoids. Addressing is in required mode.
06:15:11aaronlibot: status
06:15:23aaronlthat's a lotta factoids :o
06:15:29port 22 is ssh - SSH Remote Login Protocol
06:15:29aaronlibot: port 22
06:15:33aaronljust as i thought!
06:15:34aaronlcheating!
06:15:55BZFlag60598ED7 ?
06:16:23BZFlagaaronl: what, that's not useful to you?
06:16:30hmmm... rfc 1445 is Administrative Model for version 2 of the Simple Network Management Protocol (SNMPv2). J. Galvin & K. McCloghrie. April 1993. (Format: TXT=99443 bytes) (Status: HISTORIC)
06:16:30BZFlagibot rfc 1445
06:16:48aaronlnah, just confirming that those factoids werent all picked up from IRC
06:16:59aaronlPublic Key Server -- Error
06:16:59aaronlNo matching keys in database
06:17:03aaronl(thus i dont think so)
06:17:56BZFlagah. yep that's true. I'd like to get the "built in" ones in a different database file actually. so new ones override but not overwrite the builtins.
06:18:06aaronlyeah that would be nice
06:18:45aaronlthe key i was talking about is
06:18:45aaronlpub  1024/CD98B20A 2001/01/30 Tim Riker <Tim@Rikers.org>
06:18:47BZFlagI'm thinking of starting an ibot cvs tree as infobot seems to be lagging and I have not been able to get write access there.
06:19:01BZFlagah yes, my sig key.
06:20:17aaronlwe're connected via Joey Hess
06:20:17BZFlagoh,  oops. that is my main, the other is my Tim@Rikers.org email.
06:20:24BZFlagcool.
06:20:42anderseereading debian-private?
06:20:42aaronlit's actually an old key signature, but i just realized it a few days ago
06:20:45aaronlandersee: yeah, LOL
06:20:49anderseehehe
06:20:56aaronlthat was the best stuff i saw on that list in a long time
06:21:54BZFlaghas not been reading d-p lately.
06:22:05aaronltoday, you should
06:23:29anderseeBZFlag: worth a read this time. :)
06:25:21BZFlagthread?
06:26:03BZFlagmiller?
06:26:55aaronlyes
06:28:14BZFlagreading
06:32:20anderseeThat is just so hilarious
06:34:18aaronlonly branden would have posted that
06:34:21aaronli met him last week
06:36:55anderseehehe
06:37:11anderseeI wish I could have gone to Linuxworld, but alas it was not to be.
06:39:31aaronl:(
06:39:50aaronlbeen really busy with contract work?
06:41:54anderseeYup.  I've got a _huge_ pile of stuff I need to have done before the 1st or I will lose a government contract.
06:42:05anderseeThat contract is worth a lot of $$$ to me...
06:42:31anderseeI think I put in at least 110 hours of work this last week.
06:42:53aaronlif it's worth it, sounds great
06:44:33anderseeA pain in the butt, but yeah I think it'll be worth it.
06:44:57anderseeThe nice thing about the gov't is they actually have money, assuming you can handle the politics
08:53:59aaronlandersee: so the web site moved or something?
08:54:02aaronlno longer on your dsl?
08:57:12anderseeaaronl: The master is on my dsl line as usual.
08:57:40anderseeaaronl: But I moved busybox.net and uclibc.org onto a box belonging to a friend of mine (with few T1s)
08:58:04anderseeaaronl: Things get rsync'd onto that box from mine every few hours
08:59:02aaronlok
08:59:08aaronlis cvs.busybox.net still on the dsl?
08:59:17anderseeyup
08:59:41anderseemost folks should use busybox.net, but committers get to use my box directly.
09:00:02anderseeaaronl: did I show you this?  http://www.uwsg.indiana.edu/hypermail/linux/kernel/0208.1/1187.html
09:00:05anderseeis forgettful
09:01:44aaronljust three times :)
09:01:53aaronli showed it to you actually
09:01:53anderseehehe
09:02:18anderseeaaronl: how could you have showd it to me, when I was the one asking Al the questions?
09:02:36aaronli pointed out that he had replied
09:03:02anderseeoh
09:03:10anderseehas been working too hard
09:03:20anderseeshould take a break
09:08:15aaronldid you notice that martin resigned?
09:08:25anderseethe IDE dude?
09:08:28anderseeyeah
09:08:39anderseeHe'd been taking a lot of heat...
09:09:16anderseeAndre is about very happy about it, of course
09:15:38aaronlof course
09:15:48aaronlandre is starting to strike me as insane
09:23:59aaronlgood night
09:24:43anderseenight
14:05:20mjn3andersee: i think i found the problem with python...
14:05:36anderseeoh?
14:05:41anderseeSomething I screwed up?
14:06:20mjn3for the following python function:
14:06:25mjn3static int
14:06:25mjn3isfile(char *filename)          /* Is file, not directory */
14:06:25mjn3{
14:06:25mjn3    struct stat buf;
14:06:25mjn3    if (stat(filename, &buf) != 0)
14:06:25mjn3        return 0;
14:06:27mjn3    if (!S_ISREG(buf.st_mode))
14:06:29mjn3        return 0;
14:06:31mjn3    return 1;
14:06:33mjn3}
14:07:07mjn3i traced things... struct stat is the normal struct, but stat64 is getting called, which treats the arg as a struct stat64
14:07:34mjn3sizeof(struct stat64) = 84.  sizeof(struct stat) = 76
14:09:05anderseeHmm.  If it is calling stat64, then python was compiled with _FILE_OFFSET_BITS=64, in which case struct stat should also be getting remapped to struct stat64.
14:09:07anderseecurious
14:10:16mjn3i'll try running that file through the preprocessor
14:10:31anderseethat'd be good
14:40:09mjn3andersee: found it.  test program...
14:40:12mjn3#define _GNU_SOURCE
14:40:12mjn3#include <stdio.h>
14:40:12mjn3#include <sys/stat.h>
14:40:12mjn3int main(void)
14:40:12mjn3{
14:40:14mjn3    fprintf(stderr, "stat=%p  stat64=%p\n", stat, stat64);
14:40:15mjn3    fprintf(stderr, "sizeof(struct stat)=%d  sizeof(struct stat64)=%d\n", sizeof(struct stat), sizeof(struct stat64));
14:40:18mjn3    fprintf(stderr, "__USE_FILE_OFFSET64 is");
14:40:20mjn3#ifndef __USE_FILE_OFFSET64
14:40:22mjn3    fprintf(stderr, " not");
14:40:24mjn3#endif
14:40:26mjn3    fprintf(stderr, " defined.\n");
14:40:29mjn3    return 0;
14:40:30mjn3}
14:40:32mjn3normal gcc:
14:40:37mjn3mjn3@mars:~/work/embedded/tmp$ gcc -D_FILE_OFFSET_BITS=64 ts.c && ./a.out
14:40:38mjn3stat=0x80484f0  stat64=0x80484f0
14:40:40mjn3sizeof(struct stat)=96  sizeof(struct stat64)=96
14:40:42mjn3__USE_FILE_OFFSET64 is defined.
14:40:52mjn3uclibc:
14:40:55mjn3mjn3@mars:~/work/embedded/tmp$ PATH=~/i386-linux-uclibc/usr/bin/:$PATH gcc -D_FILE_OFFSET_BITS=64 -Wall -static ts.c && ./a.out
14:40:55mjn3stat=0x8048a4e  stat64=0x8048a4e
14:40:55mjn3sizeof(struct stat)=76  sizeof(struct stat64)=84
14:40:56mjn3__USE_FILE_OFFSET64 is defined.
14:41:03anderseeHmmm.
14:41:09anderseeLemme poke at it.
14:45:04anderseemjn3: ok, I see it.
14:58:26anderseemjn3: the problem is independent of the bits/types.h change...
14:59:39mjn3andersee: it looks to me like the padding is completely screwed up.. even before (as you say)
15:02:32anderseeyeah.  I'm trying to make some sense of it.
15:08:51anderseemjn3: bleah.  This include/bits/stat.h is stupid.  Time to rewrite it sensibly.
15:09:27anderseeI'm inclinded to through out the kernel/user struct differences here too.  This mapping is silly
15:10:00anderseethrow that is
15:10:29mjn3andersee: sounds good to me.  i was looking at it too, and can't figure out why glibc uses an unsigned long long for dev_t.
15:10:47anderseedrugs
15:11:18mjn3maybe they just hadn't reached their bloat quota that month...
15:11:19anderseeUntil Linus blesses some new plan for major/minor (or something else?) its pointless speculation
15:11:42anderseeI guess they _hope_ that whatever Linus eventually blesses will fit...
15:28:18anderseeis back (gone 18:51:37)
15:30:58mjn3andersee: looking through the kernel headers, looks like dev_t is sometimes 16 bits, sometimes 32
16:29:27tzanger'afternoon
16:29:37tzangeroyy: whereabouts in KW are you?  I'm from Listowel
16:31:13tzangerandersee: what is the securetty format that tinylogin is expecting?  I have tty1 - tty6 and ttyS0/1 listed but it refuses to let me log in as root
16:31:39tzangermjn3:  my weird sshd problems cleared up after tinylogin was installed.  perhaps it was looking for a login program, failing but still letting me through
16:33:55anderseetzanger: /etc/securetty
16:36:45tzangerandersee: yes I know...  what is the format it wants in /etc/securetty
16:37:43anderseeman 5 securetty
16:37:44anderseeNAME
16:37:44andersee       securetty - file which lists ttys from which root can log in
16:37:44anderseeDESCRIPTION
16:37:45andersee       /etc/securetty  is used by login(1); the file contains the device names of tty lines (one per
16:37:45andersee       line, without leading /dev/) on which root is allowed to login.
16:38:16tzangerandersee: ok so it requires a normal format
16:38:28tzangerand I have that
16:42:15tzangerconsole
16:42:15tzangertty1
16:42:15tzangertty2
16:42:16tzangeretc
16:44:40anderseetzanger: do you have a proper /etc/passwd, /etc/group, and if needed /etc/shadow?
16:44:54tzangeras far as I know, yes
16:45:14tzanger /bin/login will let my test user in on any tty, and ssh will let me log in as anyone
16:45:35tzangerroot:x:0:0:root:/root:/bin/sh  looks pretty straightforward to me
16:46:41anderseetzanger: ok.  And /etc/shadow?
16:47:14tzangerroot:$1$Z4leMsKx$PymD9NDQB4f36C.O0R8/u0:11914:0:99999:7:::
16:47:56tzangerone other datapoint: yes it's md5 but su works, so I'm 99% sure that tinylogin is handling md5 passwords correctly
16:53:59mjn3andersee: any progress on struct stat?
16:54:09anderseemjn3: I think I have everything ready to checkin...
16:54:21anderseeRebuilding on x86, then testing on arm
16:54:29tzangerdoes tinylogin handle securetty comments?  #somecomment lines?
16:54:33mjn3andersee: cool.  as soon as you do, i'll try python and perl again
16:54:52anderseeI've tested and seems to be working w/both 2.0.38 and 2.4.19 kernel headers.
17:04:14anderseemjn3: ok, its in CVS now...
17:06:00mjn3andersee: ok.  back in a few...
17:13:19mjn3andersee: well, python builds.. running tests and building perl now
17:14:24mjn3andersee: also, glibc stdio tests seem to run fine
17:15:22anderseeA good start.
17:15:35anderseeThese sorts of changes can be a bit scary...
17:15:35mjn3andersee: perl builds... running tests
17:15:49tzangerwhat'd you change?
17:15:56anderseewinder is _almost_ done building uClibc. :)
17:16:11anderseeI need to 2 Ghz strongArm system.
17:16:45tzangerI didn't know that strongArm went to those high clock speeds
17:16:50tzangerwhat's the power consumption like?
17:18:14mjn3andersee: perl tests completed.  now only fails the sysvipc test regarding semctl
17:18:21anderseemjn3: cool!
17:18:36mjn3andersee: all python tests passed (except linuxaudio of course)
17:18:38anderseetzanger: I was joking.  My stongarm box is 275 Mhz
17:18:44anderseenice.
17:19:03tzangerah
17:19:05anderseeCan you get an strace/ltrace of the semctl failure?
17:19:25tzangeranything weird/unusual about strongArm hw development?  (compared to dragonball, say)
17:20:27mjn3andersee: let me check...
17:20:34anderseetzanger: not too different.  You have to be careful about alignment
17:21:04tzangerits pickier about it?
17:21:26anderseeyup
17:21:48mjn3andersee: you might want to take a look at tsm.c on winder in my home dir.  it shows the bug
17:27:46tzangerhmm
17:28:01tzangerwhat does it mean "sh: can't access tty; job control turned off" ?
17:31:13tzangerhttp://www.busybox.net/lists/busybox/2002-June/011721.html
17:31:16tzangerI dont' get the joke
17:36:41tzangerhmm
17:36:50tzangerI seem to be able to put things into the background and call them back up
17:37:06tzangeroop
17:37:14tzangercan background them but cannot pull them back up
17:37:31tzangerBB_FEATURE_ASH_JOB_CONTROL is defined in the busybox config too
18:06:28tzangerscrew it
18:06:36tzangerI don't need securetty support for this anyway
18:14:15mjn3tzanger: which version of busybox are you running?
18:20:30tzanger0.60.3
18:20:45tzangerI added a patch from the mailing list to always allocate fds but it didn't hep
18:20:45tzangerer help
18:20:49mjn3tzanger: you might want to try the latest cvs version
18:21:04tzangeroh?
18:22:38tzangeryeah
18:22:39mjn3tzanger: 0.60.3 was released at the end of april.  there have been a number of bug fixes, new applets added, etc.  in fact, the devel version includes the tinylogin functionality.  i know that stuff is working because i put together a system for someone a week ago
18:22:44tzangerit was the BB_FEATURE_ALWAYS_TTY patch
18:22:54tzangermjn3: ok
18:22:58tzangerI'll give it a shot
18:23:04tzangerwho knows, maybe d<cursor> will work now too
18:23:12tzangerright now all it gives me is visual bell
18:23:18tzangercursors work fine on their own
18:24:30tzangerthe latest bb cvs snapshot?
18:25:41mjn3yes
18:25:46tzangerok
18:26:08tzangerdoes busybox's vi d<cursor> command work for you?  I'm using TERM=linux
18:27:15mjn3no idea... i didn't try it.  i was referring mainly to the login stuff.  have you tried TERM=vt100?
18:27:44tzangerno
18:28:06tzangerjust did... same result
18:28:53tzangerSUID/SGID handling is a good thing to have enabled, right?
18:31:46mjn3that's a recent addition (i think) to allow an suid root busybox.  you need that if you want busybox's su to work
18:33:16aaronloh god
18:33:20aaronlsuid busybox = scary
18:34:18mjn3andersee: the semctl problem seems to be du to the different defs of struct ipc_perm.  there is a kernel version, and then there's a lib version in bits/ipc.h with extra padding at the end
18:34:35mjn3aaronl: i thought they sorted out the suid busybox stuff
18:41:16tzangerwow
18:41:27tzangerbusybox has more stuff and yet it compiled smaller than 0.60.3
18:42:39aaronltzanger: people like me have been working on space optimizations
18:42:55tzangerI'll say
18:42:57tzanger:-)
18:50:02tzangerbah
18:50:07tzangerbb wget is no good for perl CPAN  :-)
19:01:34anderseeis away: I'm busy
19:45:05tzangermjn3 - could you test something for me with perl?
19:45:10tzangerNet::SNMP installed just fine
19:45:44tzangerthe expat library (expat.sf.net) installed fine
19:46:01tzangerXML::Parser builds fine, but all of the tests for make test fail horribly with
19:46:11tzanger /usr/bin/perl: symbol 'XmlPrologStateInit': can't resolve symbol '¬L'
19:46:21tzangersubstitute "XmlPrologStateInit" for various functions
19:46:34tzangerit seems to come down to Can't load 'blib/arch/auto/XML/Parser/Expat/Expat.so' for module XML::Parser::Expat: Unable to resolve symbol at /usr/lib/perl5/5.6.1/i686-linux/DynaLoader.pm line 206.
19:46:56tzangerI'm wondering if you get the same thing with the cvs uclibc, or if it's something I've buggered up here
20:02:54mjn3tzanger: can't resolve what symbol?
20:03:23tzanger¬L
20:03:53tzangerit's 0xa9 L IIRC
20:04:09tzangersorry 0xac
20:07:19mjn3when you say all the tests for make test fail, are you talking about for perl, or some additional packages
20:07:34tzangerall the tests for XML::Parser
20:07:47tzangerperl isntalled with only a few tests failing (3 I think, and they weren't marked as deadly important)
20:08:13tzangeralthough now that I think about it I should take a good look at the ones that fail
20:08:21tzangerthey may be directly related to XS
20:08:38mjn3yes... 2 have been fixed.  erik will be fixing the remaining one now that i've isolated it
20:09:14mjn3they shouldn't be related.  one has to do with ipc.  the others deal with passwords and groups
20:09:28mjn3s/deal/dealt/
20:09:30tzangerok well in that case no  :-)
20:11:27tzangersorry there are 6 failures
20:11:35tzangerlib/complex.t
20:11:40tzangerlib/io_multihomed.t
20:11:42tzangerlib/io_sock.t
20:11:47tzangerlib/io_udp.t
20:11:53tzangerlib/ipc_sysv.t
20:11:57tzangerop/groups.t
20:12:08tzangernone of those seemt o have anything to do with XS
20:15:36mjn3i fixed the complet failure and the groups failure.  erik is fixing the ipc_sysv failure
20:15:52tzanger*nod*
20:15:55tzangeryou guys are machines
20:15:57mjn3the io_* tests you listed are passing for me
20:16:29tzangerI am in a purely-uclibc environment if that is any difference
20:16:52mjn3that could be due to my uClibc config, or to the perl config.  don't know
20:19:06mjn3in your uClibc build dir, look for /ldso/util/readelf
20:19:41tzangerI have readelf.c, but no executable there
20:20:26mjn3do "{path to uclibc readelf}/readelf Expat.so"
20:20:32mjn3are you cross compiling?
20:20:38tzangerno cross-compile
20:20:39tzangerx86
20:21:11tzangerhmm I don't think I have a readelf binary
20:21:14mjn3just go into that directory and "make readelf"
20:21:18tzangeroh
20:21:27mjn3it doesn't look like it is built by default anymore
20:22:25tzanger# ./readelf /root/.cpan/build/XML-Parser-2.31/blib/arch/auto/XML/Parser/Expat/Expat.so
20:22:25tzangerType:           DYN (Shared object file)
20:22:25tzangerMachine:        Intel 80386
20:22:26tzangerClass:          ELF32
20:22:26tzangerData:           2's complement, little endian
20:22:26tzangerVersion:        1 (current)
20:22:28tzangerOS/ABI:         UNIX - System V
20:22:30tzangerABI Version:    0
20:22:32tzangerDependancies:
20:22:34tzanger        libexpat.so.0
20:22:36tzanger        libc.so.0
20:23:13tzanger# ./ldd /root/.cpan/build/XML-Parser-2.31/blib/arch/auto/XML/Parser/Expat/Expat.so
20:23:14tzanger        libexpat.so.0 => /usr/lib/libexpat.so.0
20:23:14tzanger        libc.so.0 => /lib/libc.so.0
20:23:29tzanger# ./ldd /usr/lib/libexpat.so.0
20:23:30tzanger        libc.so.0 => /lib/libc.so.0
20:24:31mjn3alright.  how do i go about downloading and building these packages.  i'm not a perl user
20:24:37tzangeroh
20:24:43tzangerHave you used CPAN before?
20:24:48mjn3no
20:24:55tzangerok we won't use it then :-)
20:25:12tzangerexpat.sf.net has the libexpat that XML::Parser requires
20:25:53tzangerhttp://telia.dl.sourceforge.net/sourceforge/expat/expat-1.95.4.tar.gz
20:26:39mjn3got it
20:27:08tzangerI'm trying to figure out how to download a perl module from CPAN without using the shell
20:27:15tzangerI'll just put the archive I got somewhere
20:28:41tzangerXML::Parser is at http://www.mixdown.org/XML-Parser-2.31.tar.gz
20:28:58tzangerit's your standard perl Makefile.PL / make / make test
20:49:08mjn3tzanger: seems to work for me...
20:49:11mjn3t/astress...........ok                                                      
20:49:12mjn3t/cdata.............ok                                                      
20:49:12mjn3t/decl..............ok                                                      
20:49:12mjn3t/defaulted.........ok                                                      
20:49:12mjn3t/encoding..........ok                                                      
20:49:12mjn3t/external_ent......Couldn't load LWP based external entity handler
20:49:14mjn3Switching to file-based external entity handler
20:49:16mjn3 (To avoid this message, use NoLWP option to XML::Parser)
20:49:18mjn3t/external_ent......ok                                                      
20:49:20mjn3t/file..............ok                                                      
20:49:22mjn3t/finish............ok                                                      
20:49:24mjn3t/namespaces........ok                                                      
20:49:26mjn3t/parament..........Couldn't load LWP based external entity handler
20:49:28mjn3Switching to file-based external entity handler
20:49:30mjn3 (To avoid this message, use NoLWP option to XML::Parser)
20:49:34mjn3t/parament..........ok                                                      
20:49:36mjn3t/partial...........ok                                                      
20:49:38mjn3t/skip..............ok                                                      
20:49:41mjn3t/stream............ok                                                      
20:49:42mjn3All tests successful.
20:49:44mjn3Files=13, Tests=113,  1 wallclock secs ( 0.86 cusr +  0.13 csys =  0.99 CPU)
20:52:30mjn3don't know if it makes any difference, but i have wide char support and locale support enabled.
20:56:23tzangershit
20:56:24tzangerok
20:56:33tzangerok
20:56:36tzangerI have no locale and no wchar
20:56:40tzangerI may try that and see what happens
20:56:50tzangeralthough I wonder why that would create *library* reference problems
20:58:17mjn3don't know... enabling wchar support is no problem, but enabling locale support is not for the faint-of-heart.  as soon as this latest round of debugging is finished, that's what i'm going back to working on.
20:59:23tzangerok
20:59:28tzangerI appreciate your time mjn3, thank you
21:00:06mjn3sure
21:05:22tzangerwith my luck it'll require locale since it's XML
23:05:41tzangerhmm
23:06:05tzangerI couldn't pass any tests for Compress::Bzip2 either
23:06:12tzangersame error: can't resolve symbol '¬L'
23:24:57tzangerkitchener -- whereabouts?
23:25:01tzangeris from Listowel
23:34:52anderseeis back (gone 04:33:17)
23:47:17tzangerandersee: have you ever seen weird symbol names from the linker?

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with infobot logs, split per channel and by date, etc.