**** BEGIN LOGGING AT Sun May 29 23:59:57 2005
May 30 00:05:18 <jp30> morning, jf-work
May 30 00:06:27 <CIA-8> 03jp30 * 10unslung/make/readline.mk: strip, remove static lib
May 30 00:08:30 <CIA-8> 03jp30 * 10unslung/Makefile: promote freeradius
May 30 00:11:29 <jp30> jf-work, i've been working on porting packages to wiley. what was the problem with postgresql? it builds ok for me
May 30 00:16:32 <CIA-8> 03jp30 * 10unslung/make/textutils.mk: strip textutils
May 30 00:18:28 <jf-work> jp30: dunno. I did not test it. 
May 30 00:18:49 <jf-work> btw: samba relies on openldap which still doesn't build on wiley.
May 30 00:20:15 <jf-work> and, more global question, some prerm calls /bin/killall. on wiley it is /sbin/killall. Could we put a PATH=/sbin:/bin in prerm and just call killall ?
May 30 00:20:37 <jp30> that sounds reasonable
May 30 00:21:01 <jp30> what's the status of ipkg? i don't think it's in the feed
May 30 00:21:09 <jf-work> according to you, there is no problem defining PATH in either postinst or prerm ?
May 30 00:21:39 <jf-work> I'v demote it since it causes a lot of trouble :( even on wiley
May 30 00:22:41 <jp30> re: PATH, the only danger i can think of is if you ran some daemon with a modified path, and it relied on it later
May 30 00:23:16 <jp30> re: ipkg, the other ipkg impls on wiley don't have update-alternatives, which is needed by coreutils and procps
May 30 00:23:53 <jf-work> I may ask Oleg to push the last ipkg release on its own feeds
May 30 00:23:59 <jp30> i just fixed ipkg's update-alternatives, btw - it was causing attempts to modify /usr/lib/ipkg/...
May 30 00:25:23 <jf-work> I did not understand well how ipkg works before pushing it. That was an error :(
May 30 00:27:11 <jp30> well, ipkg 0.99.148-4 is working happily on my wiley
May 30 00:27:52 <jp30> i see openldap doesn't much like uclibc
May 30 00:27:53 <jf-work> it works on me too. the trouble was at install time where it kills ipkg.conf
May 30 00:28:40 <jp30> because it replaces the existing ipkg?
May 30 00:29:09 <jf-work> openldap needs wchar_t. Oleg did not add wchar_t to uclibc and when I asked him to, he tried and got edges effects with it
May 30 00:30:27 <jp30> hmm... nasty
May 30 00:31:01 <jf-work> how is your wiley ? I got a second one this last week-end
May 30 00:31:28 <jf-work> trying to make WDS working over Radius 802.1x auth and still in trouble
May 30 00:32:33 <jp30> working nicely, it had the gx bootloader bug - i borrowed a windows laptop and fixed it today
May 30 00:33:05 <jf-work> I did it too. Is there a way to check it has been fixed ?
May 30 00:35:13 <jp30> only sure way is to compare a version number stored in nvram with the ones oleg posted to the chupa forum. ... i've just got freeradius compiling for nslu2. it turns out to be sensitive to whether various sql impls are staged
May 30 00:35:50 <jp30> on the gx bug... if your gx only seems to have 16M, then you definitely have the bug
May 30 00:36:16 <jp30> dyoung had a gx that didn't have the memory problem, but still had the switch security problem
May 30 00:36:52 <jf-work> got 32MB visible. Didn't know if I had the issue. So I did the bootloader upgrade
May 30 00:37:43 <jf-work> but I really want to have both wiley talk over WDS with either WPA or 802.1x ...
May 30 00:37:45 <jp30> well, then you're ok. i'd have been reluctant to do the upgrade if i wasn't sure i had problems. "friends don't let friends burn bootloaders w/o jtag" and all that
May 30 00:38:12 <jp30> i haven't played with WDS yet
May 30 00:41:45 <jp30> oh, i should mention, about openldap - even if the wchar_t problem can be worked around, cross-compiling it doesn't produce a working binary, even for slug
May 30 00:42:25 <jp30> caplink811 and i sweated blood trying to make it cross-compile a few months ago
May 30 00:42:37 <jf-work> are u sure ? I'have been reunning it for 1 mont on my slug
May 30 00:42:58 <jf-work> :)
May 30 00:43:31 <jp30> it's natively compiled
May 30 00:43:41 <jf-work> ah ok
May 30 00:44:32 <jp30> i take it there's no native compilation suite for wiley? crosstool doesn't seem to support mips very well
May 30 00:45:50 <jf-work> there is some uclibc buildroot available. I did not investigate on it
May 30 00:46:07 <jf-work> (coz we need a 0.9.19 uclibc which is pretty old
May 30 00:51:07 <jp30> jf-work, nice talking to you, i've gotta get some sleep now... catch you later. oh, and please do check with oleg about whether he can update ipkg to provide update-alternatives
May 30 00:51:40 <jf-work> jp30: have a good night. I will drop a mail to Oleg
May 30 00:51:56 <jp30> oh, hey one other thing, have you used "make autoclean" yet?
May 30 00:53:08 <jf-work> nop
May 30 00:53:39 <jp30> i think you might find it useful - it detects which packages have had their IPK_VERSIONs changed, and dircleans them
May 30 00:54:04 <jp30> now i really will go to sleep. 'nite
May 30 01:40:05 <dyoung> wiley online.
May 30 01:47:05 <jf-work> yeah...
May 30 01:47:25 <dyoung> Just swapped out the WRT put wiley in place.
May 30 01:47:33 <dyoung> seems to work okay.
May 30 01:49:18 <jf-work> have you more than one AP in place ?
May 30 01:50:21 <dyoung> yeah.
May 30 01:50:34 <dyoung> not talking to each other though.
May 30 01:50:36 <jf-work> good. WDS ?
May 30 01:50:51 <dyoung> I could help you figure out WDS.
May 30 01:51:00 <dyoung> I have a extra WRT now. :-)
May 30 01:51:47 <dyoung> back in 5 min, gotta reboot it, turned off upnp.
May 30 01:54:07 <dyoung> looks like it worked.
May 30 01:58:02 <jf-work> Trying to setup WDS over WEP encryption and 802.1x authentication
May 30 02:07:17 <jf-work> 802.1x auth. with EAP/TLS works fine betwwen client and wiley but both wiley cannot talk 
May 30 02:10:00 <dyoung> Hmm
May 30 02:10:21 <dyoung> Lemme check if I can configure this WRT to that level.
May 30 02:23:41 <dyoung> how do you configure this on the wiley side?  I cant seem do it through the webinterface
May 30 02:24:13 <dyoung> (woth wep AND 802.1x
May 30 02:24:14 <dyoung> )
May 30 02:26:11 <jf-work> wireless/.interface menu on the left 
May 30 02:26:32 <jf-work> you can then choose Radius with 802.1x as auth method
May 30 02:26:42 <jf-work> and WEP as encryption
May 30 02:38:18 <dyoung> what are you using for a radius server?
May 30 02:39:53 <jf-work> to authenticate end user using a pki infrastructure
May 30 02:40:23 <dyoung> yes, but I meant what program or device are you using as the radius server?
May 30 02:40:32 <jf-work> wiley
May 30 02:40:45 <jf-work> plus freeradius
May 30 02:40:49 <dyoung> Ah, okay.
May 30 02:41:03 <dyoung> I dunno how freeradius works... yet.
May 30 03:00:58 <dyoung> Hmm, I havnt some issues with making the WRT talk to the Wiley with WEP only.
May 30 03:01:04 <dyoung> so maybe I cant help you.
May 30 03:01:15 <dyoung> or maybe I'm just not lickin it right.
May 30 03:01:31 <jf-work> did you setup wds between the two ?
May 30 03:05:54 <dyoung> set a common encryption key, ssid, set the farside macs, and told them to do it.
May 30 03:06:32 <jf-work> wep or wpa-psk ?
May 30 03:08:37 <dyoung> right now just wep/
May 30 03:08:49 <dyoung> with shared keys
May 30 03:09:14 <dyoung> I have to admit though, I have only played with this once before, and it was with two identical units, so it was easy.
May 30 03:09:54 <jf-work> and the did not talk each others... I assume your using hexa keys, not passphrases
May 30 03:11:11 <dyoung> yeah.
May 30 03:11:20 <dyoung> passphrases can bite me. :-)
May 30 03:19:11 <rwhitby-away> got my zd1211 wlan usb dongle today
May 30 03:22:14 <rwhitby-away> module build under Ubuntu.  now to try and put it in OE
May 30 03:22:36 <rwhitby-away> going to use dyoung's pwc module as a template
May 30 03:22:43 <rwhitby-away> but must eat first ...
May 30 03:26:51 <dyoung> Cool.
May 30 03:26:54 <dyoung> what is this sd1211?
May 30 03:26:56 <dyoung> zd1211 ?
May 30 03:27:02 <dyoung> wireless thingee eh?
May 30 03:27:26 <rwhitby-away> yep - cheap wireless interface for the slug
May 30 03:27:34 <rwhitby-away> USD$26
May 30 03:27:51 <rwhitby-away> I got a more stylish, but more expensive version :-)
May 30 03:27:57 <dyoung> I should probably buy one.
May 30 03:28:08 <dyoung> since my Netgear WG111 seems hopeless.
May 30 03:28:40 <dyoung> jf-work, I think I'm gonna give up on WDS for today.
May 30 03:28:55 <dyoung> since I obviously dont know what the * I'm doing.
May 30 03:29:29 <jf-work> dyoung : I don't know either, but I want it to work
May 30 03:29:54 <rwhitby-away> T:  Bus=01 Lev=02 Prnt=03 Port=00 Cnt=01 Dev#=  6 Spd=480 MxCh= 0
May 30 03:29:54 <rwhitby-away> D:  Ver= 2.00 Cls=ff(vend.) Sub=ff Prot=ff MxPS=64 #Cfgs=  1
May 30 03:29:54 <rwhitby-away> P:  Vendor=6891 ProdID=a727 Rev=43.30
May 30 03:29:54 <rwhitby-away> S:  Manufacturer=3COM
May 30 03:29:54 <rwhitby-away> S:  Product=3CRUSB10075
May 30 03:29:55 <rwhitby-away> C:* #Ifs= 1 Cfg#= 1 Atr=80 MxPwr=500mA
May 30 03:29:57 <rwhitby-away> I:  If#= 0 Alt= 0 #EPs= 4 Cls=ff(vend.) Sub=00 Prot=00 Driver=(none)
May 30 03:29:59 <rwhitby-away> E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
May 30 03:30:01 <rwhitby-away> E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
May 30 03:30:03 <rwhitby-away> E:  Ad=83(I) Atr=03(Int.) MxPS=  64 Ivl=125us
May 30 03:30:05 <rwhitby-away> E:  Ad=04(O) Atr=03(Int.) MxPS=  64 Ivl=125us
May 30 03:30:14 <rwhitby-away> that's the /proc/bus/usb/devices content for it, installed on AccessSlug (OpenSlug)
May 30 03:30:23 <rwhitby-away> so it passes the smoke test ...
May 30 03:30:42 <dyoung> jf-work, I'll look at tommrow with a clear mind.  I'm sure I've missed something obvious.
May 30 03:33:07 <dyoung> rwhitby-away, does its driver have a host mode too?
May 30 03:35:41 <rwhitby-away> dyoung: I know nothing about it
May 30 03:38:49 <rwhitby-away> how would you tell?
May 30 03:43:00 <dyoung> hmm, dunno.  guess gotta read.  But it would allow you to use it as a AP.
May 30 03:48:12 <rwhitby-away> would that be part of the usb key, or host software on the linux box?
May 30 03:48:55 <dyoung> part of the driver.
May 30 03:49:08 <dyoung> the hdriver would need to know how to put it into that mode.
May 30 03:53:57 <rwhitby-away> the zd1205.c code has IW_MODE_ADHOC, IW_MODE_INFRA and IW_MODE_MASTER (which is the AP mode)
May 30 03:55:09 <rwhitby-away> so looks like it does support that
May 30 03:56:52 <dyoung> yep that all looks sorta famliiar.
May 30 03:56:56 <dyoung> Cool!
May 30 03:57:20 <dyoung> I can use one of these with AccessSlug and eliminte a AP/
May 30 03:59:18 <rwhitby-away> I would be very pleased to have a kernel module developer of your calibre on the case too :-)
May 30 03:59:52 <dyoung> heh, I was hoping you would just do it and I would just use it. ;-)
May 30 04:00:50 <rwhitby-away> hmm - I just build infrastructure and wait for others to do the work.  Looks like we need a third person to actually do the work for us ....
May 30 04:00:54 <dyoung> I'm sure it will be a easy adaptation to OE.
May 30 04:01:12 <dyoung> Yeah, I've inherieted some rwhitby qualities.
May 30 04:01:37 <dyoung> Okay point me to the source.
May 30 04:01:55 <rwhitby-away> zd1211.sf.net
May 30 04:02:08 <rwhitby-away> inheriting rwhitby qualities can't be a good thing ....
May 30 04:02:51 <dyoung> Please tell me your initial target is openslug.
May 30 04:03:04 <rwhitby-away> yep
May 30 04:03:33 <dyoung> ok I'll do it.
May 30 04:03:40 <dyoung> well lookat it at least.
May 30 04:04:04 <rwhitby-away> but the source has 2.4 kernel support too, so I might want Unslung too ...
May 30 04:19:39 <dyoung> parsing
May 30 04:20:41 <rwhitby-away> have you written the .bb file already?
May 30 04:21:06 * rwhitby-away is still getting his openslug environment back in shape so he can build pwc ...
May 30 04:24:25 <dyoung> Bah.  I need the new bitbake and berlios is down.
May 30 04:24:43 <rwhitby-away> it's on nudi
May 30 04:25:05 <rwhitby-away> ~rwhitby/openslug probably
May 30 04:25:48 <rwhitby-away> make that /home/openslug
May 30 04:26:17 <dyoung> ok
May 30 04:26:20 <rwhitby-away> are you sure berlios is down?  I got something from it just 10 minutes ago
May 30 04:26:32 <dyoung> "connection refused"
May 30 04:26:54 <rwhitby-away> svn.berlios.de ?
May 30 04:27:24 <rwhitby-away> svn+ssh
May 30 04:34:05 <dyoung> Ok, maybe I'm using the wrong url?
May 30 04:34:08 <dyoung> what is the bitbake url?
May 30 04:34:11 <dyoung> ~bitbake
May 30 04:34:29 <jbot> well, bitbake is a simple tool for the execution of tasks. It is derived from Portage, which is the package management system used by the Gentoo Linux distribution. It is most commonly used to build packages, and is used as the basis of the OpenEmbedded project.  See http://developer.berlios.de/projects/bitbake/ and http://openembedded.org/.
May 30 04:41:20 <dyoung> okay pilfered it from nudi.
May 30 04:41:26 <dyoung> bitbaking again.
May 30 04:45:52 <dyoung> guess its been a while, gotta rebuild a pile of stuffs.
May 30 04:47:16 <dyoung> summary inspection seems to indicate that its written endian agnostic.
May 30 05:12:43 <rwhitby-away> some guy on the mailing list says he's got it to work in Unslung
May 30 05:16:22 <dyoung> I'm having pebkac right now.
May 30 05:35:09 <dyoung> Ah hell.
May 30 05:35:22 <dyoung> this will compile but I think I'm gonna have to finish it up tommrow.
May 30 05:35:36 <dyoung> the Makefile needs patching beyond what I want to do right now.
May 30 05:35:49 <dyoung> you can wait that long right? ;-)
May 30 05:36:15 <dyoung> if you cant, I'll push the busted one adn someone can fix it.
May 30 05:36:28 <dyoung> I'd ratner not push broke stuff tnough
May 30 05:46:20 <rwhitby-away> I'm in no hurry
May 30 05:46:28 <dyoung> too late I think I just finished it.
May 30 05:46:30 <dyoung> heh
May 30 05:49:43 <CIA-8> 03dyoung 07 * r1.3374.1.1 10openembedded/packages/zd1211/ (zd1211_20050315.bb zd1211-20050315/Makefile.patch): Add zd1211 family of wireless usb dongle drivers
May 30 05:50:02 <dyoung> Totally untested.
May 30 05:50:17 <dyoung> I think it already has the endian patch in it though.
May 30 05:50:51 <dyoung> rwhitby-away, if you can modprobe / insmod it and it doesnt crash on load, the patches are in.
May 30 05:51:59 <dyoung> (I cheated on the makefile patch)
May 30 05:53:20 <rwhitby-away> pulling now
May 30 05:54:21 <rwhitby-away> dyoung has the potential of being the MAN here .....
May 30 05:59:54 <dyoung> I'm still thinking about how to make it work under unslung.
May 30 05:59:57 <rwhitby-away> Makefile.patch didn't apply
May 30 06:00:22 <dyoung> okay lemme look
May 30 06:01:00 <rwhitby-away> DOS file
May 30 06:01:29 <dyoung> it applied for me from a cleaned build.
May 30 06:02:29 <dyoung> strange.  it was made in linux...
May 30 06:02:39 <dyoung> I wonder why it thinks its a dos file.
May 30 06:02:46 <rwhitby-away> the Makefile has DOS line endings
May 30 06:03:22 <dyoung> do you have a easy fix for that?
May 30 06:04:03 <dyoung> cuz it works for me so I will have a hard time to diagnose/fix it...
May 30 06:05:56 <CIA-8> 03rwhitby 07 * r1.3378 10openembedded/packages/zd1211/zd1211-20050315/Makefile.patch: Added DOS line endings to the zd1211 Makefile patch
May 30 06:06:04 <rwhitby-away> done
May 30 06:06:08 <dyoung> ta.
May 30 06:07:31 <dyoung> so does it load/work?
May 30 06:09:03 <rwhitby-away> May 31 16:56:48 access user.notice kernel: zd1211 - version 5000
May 30 06:09:03 <rwhitby-away> May 31 16:56:48 access user.notice kernel: Release Ver = 3043
May 30 06:09:03 <rwhitby-away> May 31 16:56:48 access user.notice kernel: EEPORM Ver = 4330
May 30 06:09:03 <rwhitby-away> May 31 16:56:49 access user.warn kernel: usb 1-2.1: insmod timed out on ep0out
May 30 06:09:03 <rwhitby-away> May 31 16:56:49 access user.err kernel: zd1211: usb_control_msg 1 fail: FFFFFF92
May 30 06:09:05 <rwhitby-away> May 31 16:56:49 access user.err kernel: zd1211_Download_IncludeFile failed
May 30 06:09:06 <rwhitby-away> May 31 16:56:49 access user.warn kernel: zd1211: probe of 1-2.1:1.0 failed with error -5
May 30 06:09:08 <rwhitby-away> May 31 16:56:49 access user.info kernel: usbcore: registered new driver zd1211
May 30 06:09:29 <dyoung> haha in a strange twist of irony, the patch doesnt apply for me now.
May 30 06:11:18 <rwhitby-away> May 31 16:59:05 access user.info kernel: usbcore: deregistering driver zd1211
May 30 06:11:18 <rwhitby-away> May 31 16:59:18 access user.notice kernel: zd1211 - version 5000
May 30 06:11:18 <rwhitby-away> May 31 16:59:18 access user.notice kernel: Release Ver = 3043
May 30 06:11:18 <rwhitby-away> May 31 16:59:18 access user.notice kernel: EEPORM Ver = 4330
May 30 06:11:19 <rwhitby-away> May 31 16:59:19 access user.warn kernel: usb 1-2.1: insmod timed out on ep0out
May 30 06:11:19 <rwhitby-away> May 31 16:59:19 access user.err kernel: zd1211: usb_control_msg 1 fail: FFFFFF92
May 30 06:11:21 <rwhitby-away> May 31 16:59:20 access user.warn kernel: usb 1-2.1: insmod timed out on ep0out
May 30 06:11:23 <rwhitby-away> May 31 16:59:20 access user.err kernel: zd1211: usb_control_msg 1 fail: FFFFFF92
May 30 06:11:25 <rwhitby-away> May 31 16:59:20 access user.err kernel: zd1211_Download_IncludeFile failed
May 30 06:11:27 <rwhitby-away> May 31 16:59:20 access user.warn kernel: zd1211: probe of 1-2.1:1.0 failed with error -5
May 30 06:11:29 <rwhitby-away> May 31 16:59:20 access user.info kernel: usbcore: registered new driver zd1211
May 30 06:11:31 <rwhitby-away> repeatable
May 30 06:11:47 <dyoung> and lsmod doesnt show its there?
May 30 06:11:55 <rwhitby-away> yeah, lsmod shows it there
May 30 06:12:18 <rwhitby-away> Module                  Size  Used by
May 30 06:12:18 <rwhitby-away> zd1211                197936  0
May 30 06:12:41 <dyoung> http://sourceforge.net/tracker/index.php?func=detail&aid=1154655&group_id=129083&atid=713756
May 30 06:13:08 <dyoung> thats the extent of what I know about the BE patch.
May 30 06:13:46 <rwhitby-away> same here
May 30 06:13:58 <dyoung> so it doesnt quite work then?
May 30 06:14:06 <dyoung> like you cant iwconfig it?
May 30 06:14:20 <dyoung> (do we even have iwconfig?)
May 30 06:16:47 <rwhitby-away> dunno
May 30 06:17:03 * rwhitby-away has no idea about this stuff, and wonders why he even bought the thing :-)
May 30 06:18:23 <dyoung> if you ifconfig, does it show up?
May 30 06:18:31 <rwhitby-away> nope
May 30 06:18:43 <dyoung> ok its probably busted then.
May 30 06:19:10 <dyoung> oh well, valiant attempt.
May 30 06:19:12 <dyoung> :-)
May 30 06:19:24 <rwhitby-away> it's there now - someone will fix it :-)
May 30 06:19:27 <dyoung> I guess now its time to bug pjstewart about it.
May 30 06:19:51 <dyoung> I'll be more interested when I have one.
May 30 06:19:52 <rwhitby-away> I just sent him an email to the sf.net account
May 30 06:20:00 <dyoung> Cool.
May 30 06:29:51 <dyoung-zzzz> hopefully someone can fix it.
May 30 06:30:17 <rwhitby-asleep> yep
May 30 06:30:46 <dyoung-zzzz> ~emulate rwhitby
May 30 06:30:48 <jbot> Another Satisfied Customer!
May 30 06:30:59 <dyoung-zzzz> not that one!
May 30 06:31:01 <dyoung-zzzz> ~emulate rwhitby
May 30 06:31:03 <jbot> Well, my work here is done.
May 30 06:33:52 <rwhitby-asleep> dyoung-zzzz: thx for the effort
May 30 06:34:08 <dyoung-zzzz> np.
May 30 06:37:00 <mr_claus> hi
May 30 06:38:41 <mr_claus> i installed samba-3.14a and smbclient needs libreadline.so.5, which package is providing that library?
May 30 06:41:58 <jf-lunch> readline ?
May 30 06:48:07 <mr_claus> smbclient: error while loading shared libraries: libreadline.so.5: cannot open shared object file: No such file or directory
May 30 06:48:43 <mr_claus> ah, there is a package readline available
May 30 06:48:58 <jf-lunch> sorry, I should have been clearer
May 30 06:49:06 <jf-lunch> yep, try to install readline optware
May 30 06:49:16 <mr_claus> hmm, perhaps somebody can insert the dep into the samba package
May 30 06:51:35 <jf-lunch> sure
May 30 06:51:40 <CIA-8> 03jeanfabrice * 10unslung/make/samba.mk: Added readline dependency
May 30 06:53:54 <mr_claus> thx
May 30 06:53:57 <mr_claus> :)
May 30 06:54:54 <jf-lunch> np
May 30 06:55:37 <mr_claus> hmm, if i want to add libnss-ldap support to the nslu2 is it enough to build the package and configure the files in /opt/etc/ldap.conf and /opt/etc/ldap.secret?
May 30 06:55:53 <mr_claus> i changed /etc/nsswitch.conf too
May 30 06:56:11 <mr_claus> should the machine look for nsslib-ldap.so in /opt/lib?
May 30 06:56:20 <CIA-8> 03jeanfabrice * 10unslung/Makefile: Demote samba for wiley as it depends on openldap
May 30 06:57:32 <jf-lunch> shared libs have to go in /opt/lib, yes
May 30 06:57:53 <jf-lunch> if that was your question
May 30 07:03:30 <mr_claus> and it will be found by the system there? libnss_dns and libnss_files are in /lib
May 30 07:15:37 <jf-lunch> afaik, optawre also look for shared lib in /opt/lib, so yes
May 30 07:18:14 <mr_claus> hmm, ok, its not working here, grr, why software cannot only work? :/
May 30 07:19:40 <mr_claus> open("/lib/libnss_ldap.so.2", O_RDONLY) = -1 ENOENT (No such file or directory)
May 30 07:19:40 <mr_claus> open("/lib/libnss_files.so.2", O_RDONLY) = 3
May 30 07:19:52 <mr_claus> hmm, it looks in /lib instead of /opt/lib it seems
May 30 07:20:30 <jf-lunch> is it openldap taht looks in /lib ?
May 30 07:20:57 <mr_claus> i dont think so because libnss_ldap will be called before the ldap connect will be done
May 30 07:21:23 <jf-lunch> what did you strace ?
May 30 07:21:33 <mr_claus> a simple ls -l /root/x
May 30 07:22:40 <mr_claus> if i link /opt/lib/libnss_ldap-2.3.2.so -> /lib/libnss_ldap.so.2 the call need some time so i think the ldap call will be done
May 30 07:26:11 <dzlabing> hi, when i try to follow http://www.nslu2-linux.org/wiki/Puppy/BuildingPuppy I noticed that "make toolchain" fails right at the beginning - complaining about an directory not existing (because not created before), building the toolchain using "make puppy-ipk" seems to work, so i believe that there could be an error in the "toolchain" only-Makefile.  Can anyone verify this?
May 30 07:27:04 <jf-lunch> did you 'make directories' ?
May 30 07:27:50 <jf-lunch> (even if I dunno anything on "building for slug")
May 30 07:27:51 <mr_claus> hmm, if i link to /lib it ask my server, there is ldap traffic between nslu2 and the server
May 30 07:28:52 <mr_claus> now the first question is, why it's not working if the lib is in /opt/lib and second why there is no result if i link to /lib (thats a problem with my installation here i think)
May 30 07:29:24 <jp30> mr_claus, you are having trouble building with the unslung build system?
May 30 07:29:27 <jf-lunch> jp30 question :)
May 30 07:29:36 <jp30> hi jf-lunch :)
May 30 07:29:52 <mr_claus> jp30: hi, i try to integrate libnss-ldap
May 30 07:30:28 <jp30> mr_claus, that's a package seperate from openldap?
May 30 07:31:02 <mr_claus> jp30: yes, it's a libnss_ldap-2.3.2.so which is in /lib and will be called like libnss_files or libnss_dns
May 30 07:31:18 <mr_claus> jp30: it will be activated witn /etc/nsswitch.conf
May 30 07:32:11 <mr_claus> jp30: openldap itself only serve the ldap service but it doesn't really help to use samba with ldap because the filesystem need informations of ldap too
May 30 07:32:54 <mr_claus> jp30: what i want to do is to integrate the nslu2 into a pdc structure accross vpn and the ldap server is installed at the pdc
May 30 07:33:25 <jp30> does libnss_ldap actually need to be in /lib?
May 30 07:33:27 <mr_claus> jp30: now i can use that pdc to authenticate but to store files with the right permissions i need ldap informations on my nslu2
May 30 07:33:49 <mr_claus> jp30: i think libc is looking only in /lib for libnss libs perhaps
May 30 07:34:04 <mr_claus> <mr_claus> open("/lib/libnss_ldap.so.2", O_RDONLY) = -1 ENOENT (No such file or directory)
May 30 07:34:04 <mr_claus> <mr_claus> open("/lib/libnss_files.so.2", O_RDONLY) = 3
May 30 07:34:27 <jp30> i see... this looks like a similar job to gconv_modules
May 30 07:34:58 <jp30> have you looked at the glibc sources to check that this can't be configured anywhere?
May 30 07:35:10 <mr_claus> not there a two ways to solve, 1. let libc look into /opt/lib or 2. use a postinst script to link the lib into /lib
May 30 07:35:34 <jp30> (2) is what i did with gconv-modules
May 30 07:35:44 <jp30> (1) might be better if it can be done
May 30 07:35:57 <mr_claus> yes but whats happening if you upgrade your base system? you will be missing that links i think
May 30 07:36:23 <jp30> you use a startup script to put the link in again if it doesn't exist on every startup
May 30 07:36:56 <mr_claus> so the postinst script should add that information to a list of links which should be checked at startup?
May 30 07:37:17 <mr_claus> that would be the quick way i think
May 30 07:37:39 <jp30> this is the content of gconv-modules' postinst:
May 30 07:37:50 <mr_claus> i never looked into glibc, most of the time i worked with uclibc because glibc is a code monster
May 30 07:37:50 <jp30> #!/bin/sh
May 30 07:37:51 <jp30> ### establish a symlink to missing glibc bits
May 30 07:37:51 <jp30> test -e /lib/gconv || ln -sf /opt/lib/gconv /lib
May 30 07:38:17 <jp30> it also copies the same file to /opt/etc/init.d/SXXgconv
May 30 07:38:46 <mr_claus> jp30: ok, thats dirty but it will work :)
May 30 07:38:55 <jp30> well, libc on unslung is glibc, no changing that
May 30 07:39:00 <mr_claus> jp30: so you dont like to look into glibc too? :)
May 30 07:39:30 <jp30> glibc source is not great bedtime reading
May 30 07:39:39 <mr_claus> hehe
May 30 07:40:08 <jp30> jf-away, do you have a minute before you become really -away?
May 30 07:41:59 <jp30> mr_claus, we should look at glibc before using the dirty method
May 30 07:43:02 <mr_claus> jp30: i think thats a very important thing because all libs which are not basic unslung are going to /opt/lib
May 30 07:43:36 <mr_claus> whats with /etc/ld.so.conf?
May 30 07:43:55 <mr_claus> i will try to create the file and see whats happening
May 30 07:44:32 <jp30> it's important not to install anything outside /opt. manipulating outside /opt runtime is different
May 30 07:44:43 <jp30> with /etc/ld.so.conf - there isn't one. is there?
May 30 07:45:08 <mr_claus> no there is none, but perhaps it would help to have that file
May 30 07:46:25 <jp30> i've not looked at this myself, but i believe that the lack of ld.so.conf and ldconfig is a linksys legacy issue, and can't easily be fixed without breaking linksys software
May 30 07:46:33 <jp30> firmware guys would know more
May 30 07:47:32 <mr_claus> ld.so.conf makes no sense without ldconfig, hmm, i compiled the toolchain but cannot find the glibc sources in my buildtree
May 30 07:48:13 <jp30> look in toolchain/crosstool/build/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/glibc-2.2.5/nss
May 30 07:48:58 <jp30> i think the only way to configure glibc to look elsewhere for nss modules would be via ld.so.conf
May 30 07:49:36 <jp30> nsswitch.c line 340 is where it does the loading
May 30 07:50:34 <mr_claus> __libc_dlopen will be search different dirs i think
May 30 07:52:23 <jp30> looks like a neutral wrapper for dlopen to me: elf/dl-libc.c
May 30 07:54:14 <jp30> all it does is throw away errors so they don't confuse the application
May 30 07:56:35 <mr_claus> i think the right way is to use ld.so.conf
May 30 07:57:13 <jp30> well, that would be great if unslung had a working ldconfig
May 30 07:57:41 <jp30> we could try and see if there's one in the toolchain
May 30 07:58:02 <mr_claus> not there is none
May 30 07:58:28 <mr_claus> wrong, there is one
May 30 07:58:42 <mr_claus> ./armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/armv5b-softfloat-linux/sbin/ldconfig
May 30 07:59:03 <jp30> do you want to try copying it onto a slug and seeing what happens if you run it?
May 30 07:59:14 <mr_claus> yes of course
May 30 07:59:19 <mr_claus> just one moment ....
May 30 08:00:12 <mr_claus> ldconfig is 500kb
May 30 08:01:58 <mr_claus> ldconfi: Input file /lib/librt-2.2.5 not found.
May 30 08:02:00 <mr_claus> ......
May 30 08:02:05 <mr_claus> ldconfi: Input file /lib/libixml. not found.
May 30 08:02:05 <mr_claus> Segmentation fault
May 30 08:02:26 <mr_claus> thats not a native file perhaps
May 30 08:03:08 <jp30> ldconfig isn't stripped - that's why it's 500k
May 30 08:03:38 <jp30> Program received signal SIGSEGV, Segmentation fault.
May 30 08:03:38 <jp30> 0x00015ca8 in chunk_free ()
May 30 08:03:38 <jp30> (gdb) bt
May 30 08:03:38 <jp30> #0  0x00015ca8 in chunk_free ()
May 30 08:03:38 <jp30> #1  0x00015b6c in free ()
May 30 08:03:39 <jp30> #2  0x0000979c in search_dir ()
May 30 08:03:41 <jp30> #3  0x00009914 in search_dirs ()
May 30 08:03:43 <jp30> #4  0x00009e50 in main ()
May 30 08:04:46 <mr_claus> before we continue we should check if there is reason why ldconfig is not available in the firmware
May 30 08:05:40 <jp30> "it segfaults" is a reason i've heard. rwhitby and jacques might be good people to ask
May 30 08:07:47 <jp30> nothing to stop us looking more closely ourselves too
May 30 08:09:07 <mr_claus> ok, i will paste some interesting waste now
May 30 08:09:23 <mr_claus> -bash-2.05b# ./ldconfig --help
May 30 08:09:23 <mr_claus> Usag ldconfi [OPTION...]
May 30 08:09:23 <mr_claus> Configure Dynamic Linker Run Time Bindin
May 30 08:09:23 <mr_claus>   -C CACH                    Use CACHE as cache f
May 30 08:09:23 <mr_claus>   -f CONF                    Use CONF as configuration fi
May 30 08:09:24 <mr_claus>       --form=FORM            Format to use: new, old or compat (defau
May 30 08:09:30 <mr_claus>   -n                         Only process directories specified on the command
May 30 08:09:32 <mr_claus>                              line.  Don't build cac
May 30 08:09:34 <mr_claus>   -N                         Don't build cach
May 30 08:09:36 <mr_claus>   -p  --print-ca             Print ca
May 30 08:09:38 <mr_claus>   -r ROOT                    Change to and use ROOT as root directory
May 30 08:09:40 <mr_claus>   -v  --verb                 Generate verbose message
May 30 08:09:42 <mr_claus>   -X                         Don't generate links
May 30 08:09:44 <mr_claus>       --help                 Give this help l
May 30 08:09:46 <mr_claus>       --usag                 Give a short usage messa
May 30 08:09:48 <mr_claus>   -V  --vers                 Print program versio
May 30 08:09:50 <mr_claus> there are missing a lot of characters
May 30 08:09:52 <mr_claus> in every line
May 30 08:09:55 <mr_claus> perhaps its a conversion problem
May 30 08:10:12 <jp30> that does look rather garbled
May 30 08:11:29 <jp30> i've also heard that the version of glibc we're using doesn't make a working static libc. ldconfig is a static binary, ergo...
May 30 08:11:35 <mr_claus> http://www.nslu2-linux.org/wiki/Unslung/OEPackagesForUnslung
May 30 08:11:40 <mr_claus> The ldconfig executable built from OE segfaults on the slug (lots of package postinst scripts call ldconfig)
May 30 08:12:44 <mr_claus> debian sarge is using 2.3.2 too
May 30 08:13:24 <jp30> the static lib problem is an endian issue. debian arm is little endian
May 30 08:14:26 <mr_claus> ok
May 30 08:14:56 <jp30> i would like to see a ldconfig build with -g so we could get line numbers out of the gdb backtrace
May 30 08:25:22 <mr_claus> i dont know a lot about the buildsystem so it would be hard to change the flag at the right code part
May 30 08:27:33 <CIA-8> 03g2 07 * r1.3380 10openembedded/conf/distro/openslug.conf: Add jbd kernel module becase ext3 turnup needs it
May 30 08:27:36 <CIA-8> 03jbowler 07OpenSlug-1.20-pre1 * r1.3360 10openembedded/ (5 files in 4 dirs): 
May 30 08:27:36 <CIA-8> Image description changes: packages not required for a
May 30 08:27:36 <CIA-8> bootable image have been moved to the openslug.conf
May 30 08:27:36 <CIA-8> file so that an image can be built without them (by
May 30 08:27:36 <CIA-8> overriding the OPENSLUG_EXTRA_RDEPENDS variable.)
May 30 08:27:37 <CIA-8> Image release name changed to 1.20-pre1
May 30 08:31:04 <jp30> mr_claus, i'm looking at ldconfig now. i think it's safe to assume that if you want to get nss modules working with existing firmware, you'll need to use the kludgey method
May 30 08:31:24 <mr_claus> kludgey?
May 30 08:34:20 <jp30> i mean installing libnss_ldap in /opt/lib, and linking it to /lib in a startup script
May 30 08:38:18 <dzlabing> how much space is needed of a "make all" (all packages) within the unslung-CVS? 
May 30 08:42:40 <mr_claus> jp30: ok, i will see to get libnss_ldap support running
May 30 08:42:55 <jp30> dzlabing, lots...
May 30 08:45:07 <jp30> this is a guesstimate, but it wouldn't try it without 5 gig to spare
May 30 08:47:53 <jp30> mr_claus, good luck with libnss_ldap
May 30 08:48:46 <dzlabing> i've got 30gig available, but in the meantime i failed on some not available servers, 100 bit/s internet does not necassary mean to have global connectivity... but it works, after updating my tar (yes, the suse9.0 tar is outdated...)
May 30 08:55:30 <jp30> dzlabing, if an sf.net server isn't available, interrupt and restart, it will pick a different server 
May 30 08:56:24 <mr_claus> jp30: it's working
May 30 08:56:48 <mr_claus> jp30: slow but it's working, next step is to replicate the ldap part to the nslu2
May 30 08:57:24 <mr_claus> jp30: would anybody be interested in the libnss-ldap.mk if i finished the packaging?
May 30 08:57:47 <jp30> if it works, then you should contribute it
May 30 08:58:42 <mr_claus> ok, i would send the file with email
May 30 08:59:01 <mr_claus> but i have to cleanup some things and integrate the dynamic linking
May 30 09:01:35 <mr_claus> jp30: will the binary available with the standard repository?
May 30 09:04:16 <VoodooZ_Work> damn, my flash is full!
May 30 09:04:28 <VoodooZ_Work> I'll have to clean up big time.
May 30 09:04:34 <jp30> mr_claus, post the .mk file to the yahoo list, and you will get cvs write access. then you can add it to cvs, and i will push it into the feeds if i can get it to work
May 30 09:04:49 * VoodooZ_Work leaves for lunch...
May 30 09:05:22 <mr_claus> jp30: ok, i will do it if i finished it and if the file looks nice
May 30 09:33:40 <NAiL> Is berlios down, or is it just this box? (Yes, my box is acting up a bit)
May 30 09:37:28 <CIA-8> 03jp30 * 10unslung/make/usbutils.mk: usbutils: autoreconf -f does not work with automake-1.4 - use 1.9 instead, and take advantage of install-strip
May 30 09:37:55 <mr_claus> bbl
May 30 10:38:09 <macsat> hi
May 30 10:40:21 <jp30> hi macsat
May 30 10:46:02 <CIA-8> 03jp30 * 10unslung/ (4 files in 2 dirs): check in ldconfig for unslung/slug
May 30 11:08:39 <bofh_omega> hi there
May 30 11:13:39 <jp30> hi, bofh_omega
May 30 11:23:16 <CIA-8> 03jp30 * 10unslung/ (make/sed.mk sources/sed/wchar_t.patch): patch sed to compile for wiley
May 30 11:40:25 <CIA-8> 03jp30 * 10unslung/Makefile: promote ldconfig; sed and postgresql are ready for testing on wiley
May 30 11:49:54 <CIA-8> 03jp30 * 10unslung/make/usbutils.mk: usbutils: bump ipk version
May 30 12:02:04 <bofh_omega> x
May 30 12:02:27 <bofh_omega> ....and now right away reboot ;)
May 30 12:02:36 <bofh_omega> comeone...
May 30 12:02:37 <bofh_omega> ;)
May 30 12:21:41 <bofh_omega> bye
May 30 12:58:47 <jp30> hi, llagendijk. any progress with quagga?
May 30 13:02:33 <jacques> the ldd problem is because we aren't using glibc binaries which we created on unslung
May 30 13:02:40 <jacques> because when we try, they segfault
May 30 13:03:18 <jacques> so we are using the glibc binaries from the firmware dist, and some difference in our glibc build causes our ldd to not work with them
May 30 13:03:38 <jacques> ldd is tightly coupled with glibc
May 30 13:04:32 <jp30> jacques, you mean ldconfig? ldd is a shell script and works fine
May 30 13:04:46 <jacques> lol yeah that's what I meant - sorry
May 30 13:05:20 <jacques> i knew I should have let myself wake up before saying anything :-)
May 30 13:05:24 <jp30> my experiments suggest that the problem with ldconfig is only that it's a static binary, and the static lib built with the toolchain is broken
May 30 13:05:53 <jp30> the ldconfig i've packaged up is dynamically linked, so no segfault
May 30 13:10:56 <jacques> jp30, hmm interesting
May 30 13:11:35 <jacques> but I think you would find the dynamic libs are also broken
May 30 13:11:47 <jacques> tho I may be misremembering - it's been a while
May 30 13:11:55 <jp30> yes, that would make sense
May 30 13:11:58 <jacques> I do know the static is definitely broken, as you say
May 30 13:12:25 <jp30> the dynamic libs only have to be working link-time, though
May 30 13:12:34 <jacques> so you build a dynamic ldconfig and it works? fantastic!
May 30 13:12:48 <jacques> damn, why didn't I think of that :-\
May 30 13:12:48 <jp30> yeah, so far as i can tell
May 30 13:13:20 <jacques> I'm sure lots of ppl will test it - we've been needing a working ldconfig for quite a while :-)
May 30 13:13:41 * jacques goes to see if the water's boiling yet.
May 30 13:17:36 <llagendijk> jp30, yes, I am making progress, I patched configure.ac to properly detect systems without IPv6 support, but ran into some debug statements in quagga that will still try some IPv6 conversions
May 30 13:18:12 <llagendijk> jp30, I had some problems with my server today, so I spent most of the evening to get that running again
May 30 13:25:01 <jp30> llagendijk, ok, let us know when you have quagga working. good luck with the server problems
May 30 14:27:31 <CIA-8> 03jp30 * 10unslung/make/expat.mk: strip and remove static lib
May 30 15:04:57 <dyoung-zzzz> rwhitby-away, the endian patches are already in zd1211.
May 30 15:05:06 <dyoung-zzzz> just double checked it.
May 30 15:06:51 <jacques> endian patches? zd1211?
May 30 15:45:15 <dyoung-zzzz> zd1211 is rwhitby's wireless dongle.
May 30 15:45:49 <dyoung-zzzz> and on the sf page someone posted a patch for BE systems, but its already in the latest tree.
May 30 22:29:59 <Jacmet> Debian GNU/Linux 3.1 (none) ttyS0
May 30 22:30:00 <Jacmet> -bash-2.05b# uname -a                                                           
May 30 22:30:01 <Jacmet> Linux (none) 2.6.11.2 #8 Sun May 29 20:00:53 CEST 2005 armv5tel GNU/Linux      
May 30 22:30:05 <Jacmet> wee! ;)
May 30 22:41:31 <dyoung-zzzz> Cool.
May 30 22:58:00 <Jacmet> dyoung-zzzz: Now I just need to get ethernet running
May 30 22:58:26 <Jacmet> dyoung-zzzz: How's the build procedure for that? It's not in the kernel config is it?
May 30 22:58:41 <Jacmet> Me is clueless regarding the intel source
May 30 23:00:31 <dyoung-zzzz> Well, right now it wont build LE.
May 30 23:00:39 <dyoung-zzzz> thats  because only 1.4 is in OE.
May 30 23:01:04 <dyoung-zzzz> someone needs to finish working out the patches required to make 1.5 build for kernel 2.6 .
May 30 23:01:14 <dyoung-zzzz> (1.5 has LE capability)
May 30 23:01:15 <Jacmet> dyoung-zzzz: oh yes, I'll have a look at what's in oe and try to upgrade it
May 30 23:01:36 <dyoung-zzzz> I have around h alf of it done.
May 30 23:02:02 <Jacmet> nice
May 30 23:02:12 <dyoung-zzzz> but I havnt looked at it for a long time.
May 30 23:02:57 <Jacmet> dyoung-zzzz: Where is it in oe? what is it called?
May 30 23:03:29 <Jacmet> dyoung-zzzz: could you put your work up somewhere, then I might have a look at it
May 30 23:08:49 <dyoung-zzzz> I was going to, but I cant find it right now.
May 30 23:09:02 <dyoung-zzzz> its in the middle of a hard drive re-org.
May 30 23:09:14 <dyoung-zzzz> in oe, its ixp425_eth and ixp400_csr
May 30 23:09:37 <dyoung-zzzz> thats for 1.4 though, so theres no LE.  The 1.5 sources are from Intel.
May 30 23:09:46 <Jacmet> ok, thanks - I'll have a look
May 30 23:11:34 <dyoung-zzzz> you might look at beewoolies work too.
May 30 23:11:58 <dyoung-zzzz> I shamelessly pilfered a bunch of stuff from his patchset.
May 30 23:12:41 <Jacmet> ok
May 30 23:13:38 <dyoung-zzzz> http://wiki.buici.com/twiki/bin/view/Main/IXPAccessLibrary
May 30 23:14:59 <dyoung-zzzz> someone else popped in claiming to have a workable patchset and was gonna mail me patches, but I havnt heard from him.
May 30 23:15:17 <dyoung-zzzz> so you may want to google and search LAK ML for clues too.
May 30 23:15:41 <Jacmet> ok, I'll do - thanks!
May 30 23:17:38 <dyoung-zzzz> np
May 30 23:18:09 <dyoung-zzzz> (i've secretly been hoping that someone else will work out how to make CSR1.5 work, so I dont have to...)
May 30 23:21:29 <jacques> dyoung-zzzz, heh me too
May 30 23:22:10 <Jacmet> heh
**** ENDING LOGGING AT Mon May 30 23:59:57 2005