**** BEGIN LOGGING AT Thu May 05 23:59:56 2005
May 06 01:29:22 <jaro> Hi there. How I can recover wlhdd when it doesn't boot properly? By the restoration utility? (is there any possibily to run it from the linux box?)
May 06 01:30:02 <ka6sox> have you disconnected the internal HD and booted it?
May 06 01:33:05 <jaro> oh. I didn't :-) that would be probably good enough :)]
May 06 01:33:35 <ka6sox> start there and see if the firmware is hozed or just the HD.
May 06 01:35:39 <jaro> ka6sox: I was blind that i didn't see this way. The problem is, that I did set the boot_dev on that device.
May 06 01:36:43 <ka6sox> it *should* revert back to the internal jffs2 partition if the HD is missing.
May 06 01:36:48 <jaro> problem is fixed. Thanks for advice.
May 06 01:36:57 <ka6sox> np
May 06 01:37:19 <jaro> jffs2 == squashfs?
May 06 01:37:40 <ka6sox> similar...2 ways of accomplishing the same thing.
May 06 01:37:51 <ka6sox> in Slug we use jffs2
May 06 01:38:14 <jaro> two lightweight filesystems.
May 06 01:38:24 <jacques> squashfs achieves higher compression ratios but is not writable
May 06 01:38:48 <ka6sox> jffs2 has journaling.
May 06 01:39:30 <jacques> and wear-leveling
May 06 01:39:46 <jacques> neither of which matter in a ro fs :-D
May 06 01:40:00 <ka6sox> iirc the wiley's only have 4MB of Flash.
May 06 01:40:28 <rwhitby-away> yep
May 06 01:40:50 <ka6sox> so the squashfs makes sense.
May 06 01:40:54 <jaro> jacques: that would iply that squashfs is better for the purpose, isntit?
May 06 01:41:52 <jacques> define the purpose
May 06 01:42:15 <jacques> if rw is not needed then yes
May 06 01:44:33 <rwhitby-away> Oleg chose LZMA squashfs for Wiley cause he only has 4MB to work with, and it would be over that if he converted to JFFS2 now.
May 06 01:45:14 <ka6sox> now replacing the Flash is a real PITA.....5mm pitch.
May 06 01:45:35 <ka6sox> er thats 0.5mm pitch.
May 06 01:46:43 <ka6sox> rwhitby-away, does the WL500g/dlx only have one RAM chip?
May 06 01:46:52 <jaro> jacques: maybe Slug is not such limited in flashspace, but asus routers, as rwithby mentioned are. The purpose in general might be reformulated as: ro fs on really small flash.
May 06 01:48:49 <jacques> rwhitby-away, what's taking all the space?
May 06 01:50:12 <jacques> is it like the slug where samba stuff is taking several MB it doesn't really need?
May 06 01:51:08 <jacques> I know it's possible to fit bootloader, kernel, uclibc, busybox, X libs, X server, window manager, rxvt, and more stuff in 4MB jffs2 - we did it on the tuxscreen
May 06 01:53:25 <ka6sox> thats a lot o' stuff to cram into that little space.
May 06 01:53:45 <jacques> well there wasn't much left after that :-)
May 06 01:53:56 <jacques> and of course would not have been possible with glibc
May 06 01:55:44 <jacques> I just dusted off a couple of my tuxscreens the other night and plugged them in
May 06 01:55:47 <rwhitby-away> jacques: wl500gx has a lot more features built-in than the slug.  e.g. webcam support including motion detection ....
May 06 01:56:03 <jacques> rwhitby-away, ah ok I start to understand
May 06 01:56:18 <jacques> it probably has lots of driver modules
May 06 01:58:45 <jaro> however a lot of things on asus aren't usable in every piece :)
May 06 01:59:25 <rwhitby-away> jaro: what do you mean?  Oleg has #defines for various hardware capabilities ...
May 06 02:03:44 <jaro> rwhitby: I mean, that lot of stuff (e.g. webcam) can be used but not everywhere. I didn't say, that they can't be excluded in custom firmware - I do that massively :)
May 06 02:04:56 <jaro> this is the case when developing special purpose firmwares, where just basic capabilities and bunch of own sw are required.
May 06 02:16:27 <rwhitby-away> yep, you can do that with the ,config file for Oleg's software ...
May 06 02:17:13 <rwhitby-away> Asus have to put in all the drivers for the superset of features that can be used on that hardware platform (e.g. webcam on any of them that have USB).
May 06 02:38:10 <dyoung> lennert_zzz, I didnt make the post today.  I'll go tommrow afternoon.
May 06 02:40:59 <rwhitby-away> dyoung: loaded Unslung 4.19-alpha yet?
May 06 02:42:37 <dyoung> Nope.  Been busy with RL stuff.  I hit a milestone today though and was thinking I'd foolwith it now; since I freed a extra slug.
May 06 02:43:35 <dyoung> dl-ing now
May 06 03:07:43 <macsat> hi guys
May 06 03:11:21 <phedny> morning, macsat :)
May 06 03:11:58 <macsat> anyone in here with cross compiler experience, that can give me a few hints on how to "get started" ?
May 06 03:12:18 <macsat> I have already the build stuff in order in my linux (incl. toolchain)
May 06 03:28:30 <dyoung> what are you trying to do, on what plaform?
May 06 03:30:21 <jaro> Having troubles with wlhdd. I'm trying to have rootfs on internal disk. However, don't know, what's wrong. When I did set boot_dev it didn't work properly. So I tried to manually perform pivot_root ; exec chroot . sbin/init. In that moment, router has disconnected me. However httpd & telnetd are running. Just telnet is refusing connections.
May 06 03:41:47 <macsat> dyoung, I am simply trying to figure out how to create .ipk packages. My desktop computer is a Fedora Core 3. I have followed instructions on how to setup the build environment...now I need to find out how to actuately compile something as a ipk package....just something simple at first, then I can always move on to getting more complex :)
May 06 03:41:58 <macsat> Im simply trying to figure out HOW to do it
May 06 03:42:26 <macsat> It is for the Wiley platform btw, but that shouldnt make much of a difference I guess....trying to help out jf a bit :-)
May 06 03:53:03 <jacques> macsat, you've looked at the wiki, pages like this: http://www.nslu2-linux.org/wiki/Unslung/AddAPackageToUnslung ?
May 06 03:53:42 <macsat> yeah....I fear I am too stupid to get it ;-)
May 06 03:57:14 <jacques> do you know how Makefiles work?
May 06 04:11:24 <rwhitby-away> macsat: we can help you here.  Have you checked out the Unslung package build environment from the CVS as described in the page that jacques referenced?
May 06 04:12:13 <macsat> yes rwhitby, that I have.
May 06 04:12:21 <rwhitby> your first step is to be able to build the existing packages for the wl500g.
May 06 04:12:36 <macsat> yeah, I kindda figured that would be a good place to start :)
May 06 04:12:43 <rwhitby> so you have an "unslung" directory, with a top-level Makefile, and "make" and "sources" subdirs?
May 06 04:12:47 <rwhitby> (among others)
May 06 04:12:55 <macsat> I have downloaded the asus tar ball, and also updated with olegs toolchain, so...
May 06 04:12:59 <macsat> yes
May 06 04:13:17 <macsat> and I did : make directories   and   make ipkg-utils
May 06 04:13:18 <rwhitby> the "unslung" directory I am referring to is a different thing from the toolchain
May 06 04:13:25 <rwhitby> ok, good.
May 06 04:13:26 <macsat> yes, I know :)
May 06 04:13:45 <rwhitby> have you changed UNSLUNG_TARGET to wl500g in the top-level Makefile?
May 06 04:13:54 <macsat> nope....
May 06 04:13:55 <macsat> :)
May 06 04:13:55 <rwhitby> (by default it is nslu2)
May 06 04:14:01 <rwhitby> ok, do that first.
May 06 04:14:15 <rwhitby> BTW, are you going to write up this in one of your excellent howto's ?
May 06 04:14:18 <macsat> now I have
May 06 04:14:28 <rwhitby> ok, now type "make"
May 06 04:14:32 <macsat> yes, I planned to rwhitby  :)
May 06 04:14:43 <rwhitby> it should start building all the wl500g packages
May 06 04:15:07 <rwhitby> it uses the external toolchain that you have installed in /opt/brcm
May 06 04:15:26 <rwhitby> which must have Oleg's additions, as you have also done
May 06 04:15:46 <macsat> yes
May 06 04:15:53 <macsat> but it gives me a few errors
May 06 04:16:00 <rwhitby> ok, what's the first error.
May 06 04:16:23 <macsat> Unreal3.2.3.tar.gz
May 06 04:16:26 <macsat> ups
May 06 04:16:41 <macsat> First error : bzcat: Can't open input file /opt/unslung/downloads/busybox-1.00.tar.bz2: No such file or directory.
May 06 04:17:06 <rwhitby> ok, when did you do cvs update -dP last ?
May 06 04:17:23 <macsat> sure enough, in my /opt/unslung/downloads   dir I only have ipkg-utils-1.7.tar.gz
May 06 04:17:28 <macsat> maby I missed something :Å)
May 06 04:17:36 <rwhitby> that was a real error that jp30 fixed today
May 06 04:17:45 <macsat> ok
May 06 04:17:56 <rwhitby> BTW, what type of machine and OS are you doing this building on?
May 06 04:18:03 <macsat> fedora core 3
May 06 04:18:05 <dyoung> man that took a long time.
May 06 04:18:09 <macsat> i586
May 06 04:18:22 <macsat> or i686 or something
May 06 04:18:23 <macsat> :)
May 06 04:18:28 <dyoung> unsling hdd-conf where hdd is a 256MB Flash device took almost 30 miin.
May 06 04:18:42 <rwhitby> macsat: that should be fine.
May 06 04:18:51 <macsat> where do I do the cvs update -dP    ?
May 06 04:18:59 <jacques> in the unslung dir
May 06 04:19:08 <rwhitby> do a "cvs update -dP" now, to make sure you have the latest.  Just inside the unslung directory (which has the top-level Makefile"
May 06 04:19:13 <macsat> hmm...the only cvs I have done, is actuately : cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/nslu co unslung
May 06 04:19:20 <macsat> So I guess I have been a little too fast :)
May 06 04:19:22 <macsat> will do
May 06 04:19:54 <macsat> updating :)
May 06 04:20:02 <macsat> ok, done
May 06 04:20:08 <macsat> should I try to 'make' command again ?
May 06 04:20:18 <rwhitby> ok, which package was it trying to build when it got an error ?
May 06 04:20:25 <macsat> yeah, seems better now :)
May 06 04:20:39 <rwhitby> ok, so it's building packages now?
May 06 04:20:45 <macsat> it was busybox, this time it is downloading the tar.gz instead though. Seems more reasonable
May 06 04:20:54 <macsat> yeah, downloading, and now compiling
May 06 04:21:09 <rwhitby> ok, good.  First step almost complete.
May 06 04:21:18 <macsat> I guess this could take a while, building all packages, since I am only on a 512/512 connection
May 06 04:21:33 <macsat> and my CPU is only an Athlon XP 2100+ :-)
May 06 04:21:36 <rwhitby> ok, which package are you most interested in?
May 06 04:21:42 <rwhitby> and we will concentrate on that
May 06 04:22:07 <macsat> You mean which of the existing ones I would like to try to build ?
May 06 04:22:12 <rwhitby> yes
May 06 04:22:28 <rwhitby> note that you might also need to install some of the tools mentioned in the Debian instructions on this page: http://www.nslu2-linux.org/wiki/OpenEmbedded/RequiredSoftware
May 06 04:22:35 <macsat> dunno...maby 'less' would be a good choice for starts
May 06 04:22:54 <rwhitby> ok, do "make less-ipk"
May 06 04:23:13 <macsat> It is not done, I guess it is doing it all, since I only wrote 'make'
May 06 04:23:20 <macsat> shouldnt I just let it run ?
May 06 04:23:25 <rwhitby> no, just Ctrl-C
May 06 04:23:28 <macsat> ok
May 06 04:23:43 <macsat> doing less now
May 06 04:23:43 <rwhitby> actually, I will take you through it step by step.
May 06 04:24:29 <rwhitby> ok, bring up make/less.mk in a viewer or editor window so I can explain any questions you have about that
May 06 04:24:34 <macsat> checking for working terminal libraries... Cannot find terminal libraries - configure failed
May 06 04:24:46 <macsat> ok, lets do the less.mk thing
May 06 04:24:53 <macsat> I can always find the required soft after that :)
May 06 04:25:07 <macsat> less.mk is loaded in emacs :)
May 06 04:26:20 <rwhitby> ok, note the definitions at the top, then the download (less-source) target, then the configure target (.configured), then the build target "$(LESS_BUILD_DIR)/less", then the control file target, then the -ipk target
May 06 04:26:31 <rwhitby> that's the main structure, and main flow of execution
May 06 04:26:42 <macsat> ok
May 06 04:26:53 <rwhitby> and less is a good example of a normal GNU autotools-style package
May 06 04:27:09 <macsat> great :)
May 06 04:27:41 <rwhitby> we run "configure" with the right options in the .configured target, then we run make with the right options to actually build the executable, then we package it up (generating the control file in the process from the definitions at the top).
May 06 04:27:44 <macsat> dosnt look too hard
May 06 04:27:59 <rwhitby> A simple package can be done in less than five minutes from download to feed
May 06 04:28:02 <macsat> how do I know which options are the "right" ?
May 06 04:28:13 <rwhitby> template.mk is your friend
May 06 04:28:21 <macsat> ok
May 06 04:28:49 <macsat> so most packages that are not too complex, actuately builds with only a few special options configures at the top ?
May 06 04:28:54 <rwhitby> yes
May 06 04:29:08 <rwhitby> but others are incredibly complex
May 06 04:29:17 <rwhitby> (like the ones that jp30 does)
May 06 04:29:20 <macsat> I can imagine
May 06 04:29:33 <rwhitby> Now, note the LESS_DEPENDS definition at the top.
May 06 04:29:40 <rwhitby> that is a run-time dependency on another package.
May 06 04:29:42 <macsat> yes...ncurses
May 06 04:30:01 <macsat> how does one know this? 
May 06 04:30:32 <rwhitby> well, you found it just before - the package wouldn't build, cause it was looking for terminal functions (ncurses)
May 06 04:30:41 <rwhitby> you have actually found an error in less.mk
May 06 04:31:02 <rwhitby> it is missing a call to "make ncurses-stage" in the .configured target.
May 06 04:31:06 <macsat> ok, since it didnt tell me ncurses
May 06 04:31:07 <macsat> ok
May 06 04:31:13 <rwhitby> Do you have an sf.net id?
May 06 04:31:17 <macsat> yeah
May 06 04:31:24 <rwhitby> what is it?
May 06 04:31:26 <macsat> ...at least I USED to ;-)
May 06 04:31:29 <macsat> I belive it is macsat
May 06 04:31:41 <macsat> will check, havent used it in a while....w8
May 06 04:32:25 <macsat> sure enough, macsat it is :)
May 06 04:32:43 <rwhitby> do you have an ssh key loaded in sf.net?
May 06 04:33:17 <macsat> nop
May 06 04:33:35 <rwhitby> because you are now an nslu2-linux packages developer, so you will need one to be able to write to the CVS repository
May 06 04:33:56 <rwhitby> ok, edit less.mk and add a call to "make ncurses-stage" as the first line of the .configured target
May 06 04:34:00 <rwhitby> then run "make less"
May 06 04:34:11 <macsat> ok, will try...w8
May 06 04:34:51 <rwhitby> the ncurses-stage target tells ncurses.mk to put it's include files and library files in the .../staging/opt/... directories, so that less.mk can find them there
May 06 04:35:24 <rwhitby> because we are cross-compiling, we must always get headers and libs from staging, instead of from /usr/... on the host build machine
May 06 04:35:40 <macsat> this seems so much better :)
May 06 04:36:09 <macsat> yeah, I can imagine that a i386 ncurses isnt really usable :)
May 06 04:36:46 <CIA-1> 03rwhitby * 10unslung/make/less.mk: Added the call to ncurses-stage.
May 06 04:36:47 <macsat> make.3-80 compiled, and now it is compiling ncurses
May 06 04:36:56 <rwhitby> let me know if I'm explaining things that you already know.
May 06 04:37:06 <rwhitby> or if I'm pitching things at the right level
May 06 04:37:09 <macsat> make must de defined as needed in ncurses
May 06 04:37:19 <macsat> you are not. I have done compiling, but never cross compiling
May 06 04:37:33 <rwhitby> ok, that was what I was assuming
May 06 04:37:37 <macsat> not much of a programmer. I can READ some code, and write a bit as well....but not much :)
May 06 04:38:10 <rwhitby> you are one hell of a documenter though ....
May 06 04:38:18 <macsat> thanks :)
May 06 04:38:39 <rwhitby> ok, let me know when less has finished compiling
May 06 04:38:42 <macsat> trying to do it in a way, so even non-linux-knowable people can use it....even though that is hard sometimes.
May 06 04:39:07 <jeanfabrice> 'morning all
May 06 04:39:13 <rwhitby> jeanfabrice: g;day
May 06 04:39:17 <macsat> when you have been using linux for some 10 years, it can be hard not to take some things as granted :)
May 06 04:39:19 <macsat> hi jeanfabrice 
May 06 04:39:32 <rwhitby> macsat is becoming an Optware developer ...
May 06 04:39:41 <jeanfabrice> that's what I see ;)
May 06 04:40:08 <jeanfabrice> Have you done your wileyware upgrade ?
May 06 04:40:33 <macsat> I havent yet...still on -4
May 06 04:40:45 <rwhitby> me neither - I plan to do it later tonight
May 06 04:41:03 <macsat> ok - less is done, and less_381-2_mipsel.ipk is created :)
May 06 04:41:19 <macsat> maby I should try it out on my box...
May 06 04:41:31 <jeanfabrice> was a pain to recompile. even is Makefile is more clear than ever
May 06 04:41:49 <rwhitby> I'm soldering up some dual power cables so I can run both the nslu2 and a 4 port hub off the same power pack.  Same for the wl500gx and a 7 port hub.  Trying to cut down on the number of power packs in the wall.
May 06 04:42:09 <rwhitby> So I plan to do the firmware upgrade when I take the internet offline to replace the power to the wl500gx
May 06 04:42:52 <rwhitby> macsat: you can scp that .ipk file up to /tmp/harddisk, and then run "ipkg install /tmp/harddisk/less_*.ipk" to install it
May 06 04:43:01 <jeanfabrice> the adapter you use is one of the two ? 
May 06 04:43:04 <macsat> yepp...will try
May 06 04:43:44 <macsat> scp ?...sounds like something I dont know about....I have just ftp'ed it up there now ;)
May 06 04:44:03 <rwhitby> jeanfabrice: because most of my stuff has come from the US, I have local power packs (most are 2.5A) for powering everything.  The slugs, wileys and hubs are all 5V, so i can run multiple things off the one power pack
May 06 04:44:10 <macsat> WORKING  :-)
May 06 04:44:18 <macsat> nice, thanks a lot rwhitby 
May 06 04:44:26 <macsat> seems like my "develop install" works 
May 06 04:44:41 <macsat> so, now it is "just" a question of figuring out how to create my own packages
May 06 04:44:51 <rwhitby> macsat: I consider this an investment, as I know that you will be helping many more Optware/Wiley developers get their feet ...
May 06 04:44:54 <macsat> I will try to find something fairly simple
May 06 04:45:23 <macsat> yeah, I will definately try to give a hand....I'm here for a reason you know :)
May 06 04:45:32 <rwhitby> the other big thing to know about is the sources directory, where patches and rc files and postinsts and the like are stored
May 06 04:46:02 <rwhitby> less has no patches required, and no rc scripts or postinst or prerm files, so it has no directory in sources.
May 06 04:46:11 <macsat> ok
May 06 04:46:25 <jeanfabrice> rwhitby: I was wondering also how to reduce the number of powerpack on the wall. I did not find powerpack with such 2.5A. I can easily find 800mA
May 06 04:46:26 <macsat> so only patches and stuff goes in sources, not the real source-code for the software ?
May 06 04:46:46 <rwhitby> correct.  The real sourcecode is always downloaded
May 06 04:47:15 <macsat> ok
May 06 04:47:55 <rwhitby> ok, run "make dnsmasq-ipk"
May 06 04:48:11 <rwhitby> and look in dnsmasq.mk and sources/dnsmasq/*
May 06 04:48:22 <macsat> so if I decide to try to find a package that is not directly downloadable, but only through some stupid cgi / php script that masks the real URL, I will need to put the source somewhere on the web where it can be accessed ?
May 06 04:48:28 <macsat> ok
May 06 04:48:41 <rwhitby> yes, and we can do that in nslu.sf.net/downloads
May 06 04:48:46 <macsat> ok
May 06 04:49:00 <macsat> done already...that was a fast one
May 06 04:49:16 <rwhitby> ok, notice the DNSMASQ_PATCHES line in dnsmasq.mk
May 06 04:49:31 <rwhitby> and the third command in .configured which applies those patches
May 06 04:49:35 <macsat> yeah...
May 06 04:49:59 <rwhitby> if something can't be done with options to configure or make, then we have to patch it like that
May 06 04:50:22 <dyoung> Okay I'm unslunged to 4.19 now.  I did it serial-less from 3.17; unslinging to the same partition that had the 3.17 /opt in it.
May 06 04:50:26 <dyoung> worked seamless.
May 06 04:50:29 <rwhitby> macsat: you know how to use diff to create patches?
May 06 04:50:39 <macsat> would say :)
May 06 04:50:39 <dyoung> It just took a really long time.
May 06 04:50:42 <macsat> wouldnt I mean
May 06 04:50:50 <macsat> I have APPLIED patches before, but never created
May 06 04:51:29 <macsat> I get it thought I think
May 06 04:51:41 <macsat> - = old line in source, + = new line
May 06 04:51:54 <macsat> from line 40 -> 7 lines forth in source
May 06 04:52:20 <rwhitby> ok, have a read of the "patch" and "diff" man pages later.  Remember to either use "diff -c" or "diff -u" to create the patches, and make sure you start from the builds dir to do it (so the -p1 argument to patch is correct)
May 06 04:52:42 <macsat> yeah, I will. Once the "need" occur :)
May 06 04:52:47 <rwhitby> next thing to note is the postinst and prerm files in sources/dnsmasq
May 06 04:53:06 <macsat> will most software need patches, in order to store conf files and so on in /opt  instead of /
May 06 04:53:07 <macsat> ?
May 06 04:53:22 <rwhitby> usually a "--prefix=/opt" to configure will do it
May 06 04:53:49 <jf-away> rwhitby: will you echoing irclogs > /web/howto ?
May 06 04:54:01 <rwhitby> or sometimes a "prefix=/opt" or "PREFIX=/opt" in the Makefile
May 06 04:54:09 <rwhitby> jf-away: macsat is going to write up a nice howto for us
May 06 04:54:33 <jf-away> good macsat !
May 06 04:54:50 <macsat> I will write a howto on how to get the build-stuff running on your PC, and how to test your packages....
May 06 04:55:02 <macsat> then maby later I will try to add a howto in how to create own packages
May 06 04:55:12 <rwhitby> excellent.
May 06 04:55:15 <macsat> I need to be a little more confident there first :)
May 06 04:55:51 <rwhitby> now, also note the rc.dnsmasq file - this is the startup script which ends up in /opt/etc/init.d/SXXdnsmasq
May 06 04:56:00 <macsat> I will save the log of this channel for future reference a swell ;-)
May 06 04:56:00 <rwhitby> (where XX is a number which determines the startup order)
May 06 04:56:06 <macsat> yeah, saw it
May 06 04:56:15 <macsat> how does one determine the XX ?
May 06 04:56:31 <rwhitby> very informally :-)
May 06 04:56:35 <macsat> hehe ok
May 06 04:56:42 <macsat> good enough
May 06 04:56:56 <dyoung> Roll the 10-sided die twice?
May 06 04:57:12 <macsat> dosnt even matter if 2 are identical either....S11abc will just come before S11bcd if that happens :)
May 06 04:57:18 <rwhitby> right
May 06 04:57:44 <rwhitby> one way is to see what it has on a desktop system
May 06 04:58:04 <macsat> yeah, could be a reasonable way to figure it out.
May 06 04:58:30 <rwhitby> it only really matters when there are dependencies in package startup ordering
May 06 04:58:43 <macsat> does most of the packages in the make  dir compile on wiley as well ?
May 06 04:58:56 <macsat> I mean work, not just compile ;)
May 06 04:58:58 <rwhitby> ok, look at the top-level Makefile again
May 06 04:59:02 <macsat> ok
May 06 04:59:14 <macsat> looking at it
May 06 04:59:19 <rwhitby> WL500G_PACKAGES and following definitions
May 06 05:00:03 <rwhitby> one way you can get your feet wet in package development is to fix some of the WL500G_PACKAGES_JUST_REQUIRING_CONTROL_GENERATION
May 06 05:00:35 <rwhitby> early on, we used a static control file.  Then jp30 worked out how to generate them.  All wiley packages must have generated control files.
May 06 05:00:49 <macsat> control files beeing ?
May 06 05:01:05 <rwhitby> the file which has all the information for ipkg to use
May 06 05:01:38 <rwhitby> look for builds/dnsmasq-*-ipk/CONTROL/control for an example
May 06 05:01:58 <rwhitby> it now gets generated from definitions at the top of foo.mk
May 06 05:02:52 <rwhitby> to convert a package, you just need to add those definitions, add the target which generates the control file, and replace the command in the ipkg target which used to copy the static control file from the sources dir with a call to make to generate the new control file
May 06 05:03:12 <rwhitby> then you increment the IPK version, and you've made a contribution to the community :-)
May 06 05:04:03 <rwhitby> Move the package name to the WL500G_PACKAGES_READY_FOR_TESTING definition, and let jeanfabrice know.  He will build it on the official build machine, and upload it to the feed.
May 06 05:05:32 <rwhitby> now, as you stated before, just cause a package builds, doesn't mean it runs.  We also need people to test out all the packages listed for the WL500g, cause we haven't tested that many of them
May 06 05:05:56 <rwhitby> As Linus Torvalds says "Testing is for users"
May 06 05:07:22 <rwhitby> and we need to be careful that all the dependencies for a wl500g package are also wl500g packages.
May 06 05:08:43 <macsat> I will give it a try :)
May 06 05:09:28 <rwhitby> have a read of the sf.net docs for how to check out the cvs repo as a developer (using :ext: instead of :pserver:) and uploading your ssh keys so you can write back.
May 06 05:10:17 <macsat> She-Who-Cant-Be-Ignored has just returned home.....and today beeing her birthday and all, I better pay her a little attention as well ;-)
May 06 05:10:25 <rwhitby> then there are 11 packages needing control file generation ready for you :-)
May 06 05:10:31 <macsat> I will...
May 06 05:10:40 <rwhitby> happy birthday her from the nslu2-linux crew!
May 06 05:10:59 <rwhitby> and good luck with the packages.
May 06 05:11:25 <rwhitby> now go be with your family :-)
May 06 05:11:42 <dyoung> speaking of testing... anyone know if ka6sox or anyone else has tested the audio packages I ci'd couple months ago?  vorbis stuff and friends and devices and misc other stuff ?
May 06 05:12:39 <macsat> so I will need to do a new kind of download from the CVS ?
May 06 05:12:57 <macsat> I just bought a few minutes more, she needed to talk to her sister ;-)
May 06 05:13:10 * rwhitby goes back to the soldering ...
May 06 05:13:41 <dyoung> I need hotair.
May 06 05:14:15 <[g2]> hey what did jb change the OpenSlug pw to ?
May 06 05:14:16 <rwhitby> macsat: yes, no need to either re-download as :ext: or do some sedding on your CVS/Root files throughout your checked out dir
May 06 05:14:32 <macsat> okidoki - redownloading is probably the easiest :)
May 06 05:14:46 <rwhitby> macsat: get your ssh key set up first
May 06 05:14:50 * [g2] has been password-challenged that last couple days :)
May 06 05:14:54 <macsat> yeah, I will
May 06 05:15:01 <[g2]> morning all
May 06 05:15:04 <dyoung> opeNSLUg ?
May 06 05:15:12 <macsat> Will look into all this. Thanks for helping thus far rwhitby  :)
May 06 05:15:20 <rwhitby> macsat: np
May 06 05:15:38 <rwhitby> [g2]: it's either "opeNSLUg" or "OpenSlug"
May 06 05:16:09 <rwhitby> "opeNSLUg" according to the README in base-passwd/openslug
May 06 05:16:31 <dyoung> I'm nearly positive we voted for opeNSLUg
May 06 05:18:02 <NAiL> Worked with opeNSLUg last time I built
May 06 05:18:15 <[g2]> it's definetly not OpenSlug that was my init mod and it stopped working after the fresh build from last night
May 06 05:18:37 <[g2]> opeNSLUg is probably it... We'll know in a couple of seconds
May 06 05:18:49 <[g2]> bingo
May 06 05:18:58 * [g2] does the happy dance again
May 06 05:19:01 <[g2]> THX all
May 06 05:19:20 <rwhitby> is the linuxrc working now?
May 06 05:19:34 <[g2]> that's what I was testing
May 06 05:19:47 <[g2]> the build from late last night with jb's checkin
May 06 05:19:49 <[g2]> s
May 06 05:19:58 <rwhitby> I wasn't able to test it here cause I still don't have a spare LAN port
May 06 05:20:17 <[g2]> sounds like you need an RV082 :)
May 06 05:20:30 <rwhitby> not at that price!
May 06 05:21:11 <[g2]> in the not too distant future I'll be doing native compiles on it @ 533
May 06 05:21:25 <rwhitby> native schmative
May 06 05:21:36 <NAiL> [g2]: I  take it I can safely flash my slug as well now? ;)
May 06 05:21:57 <[g2]> let me try ssh'ing in
May 06 05:22:32 <NAiL> [g2]: You built with glibc, right?
May 06 05:22:39 <[g2]> nod.
May 06 05:22:42 <[g2]> I'm in
May 06 05:23:04 <[g2]> NAiL, I haven't vegtabled yet
May 06 05:23:12 <[g2]> but the jffs2 is up
May 06 05:23:35 * rwhitby just finishes compiling Oleg's latest WileyWare
May 06 05:24:08 <[g2]> I've always thought the wl-500xxx seemed like a neat toy
May 06 05:24:16 <[g2]> s/neat toy/cool device/
May 06 05:24:21 <NAiL> [g2]: Vegtable?
May 06 05:24:30 <rwhitby> turnup
May 06 05:24:35 <dyoung> ~nslu2-veggie
May 06 05:24:44 <[g2]> these guys always tease me about that name :)
May 06 05:24:52 <NAiL> hehehe
May 06 05:25:55 <jacques> neat, I just got binfmt_misc module working
May 06 05:26:14 <[g2]> rw-soldering, what are you soldering ?
May 06 05:26:24 <jacques> so my kernel knows to automatically use qemu to run armeb binaries
May 06 05:26:39 <jacques> I just type the binary name as if it were x86 an it runs :-)
May 06 05:26:52 <[g2]> jacques, that sounds cool
May 06 05:27:15 <[g2]> I've been wanting to play with qemu and the armeb binaries 
May 06 05:27:23 <[g2]> you're always so far ahead!
May 06 05:28:08 <[g2]> can we compile qemu for the Xscale then just run all the x86 binaries from the slug ? :)
May 06 05:28:28 <dyoung> interesting concept.
May 06 05:28:33 <NAiL> uh.. slug is on x.x.1.77,  right? 
May 06 05:28:42 <[g2]> mine was
May 06 05:28:44 <jacques> [g2], I considered that
May 06 05:28:48 <dyoung> that would probably make appweb configurator work.
May 06 05:28:50 <jacques> but I don't think we have the ram
May 06 05:29:32 <[g2]> jacques, with help from dyoung and robbat2 I got NFS swap running on the slug early today
May 06 05:30:43 <jacques> [g2], wow, NFS swap works now?
May 06 05:31:02 <[g2]> well the way I setup it up it did
May 06 05:31:09 <[g2]> it's pretty trivial
May 06 05:31:16 <[g2]> I'll write it up in the wiki today
May 06 05:31:31 * [g2] wanders over to the wiki right now
May 06 05:31:31 <jacques> this reminds me of when I built/ran bochs on my ipaq
May 06 05:31:38 <jacques> I booted x86 linux on it
May 06 05:31:45 <[g2]> that's just funny
May 06 05:31:58 <jacques> some screenshots here: http://intimate.handhelds.org/screens.html
May 06 05:32:00 <jacques> this was in 2001
May 06 05:32:37 <jacques> I also booted DOS on it :-)
May 06 05:32:43 <[g2]> so do you think I should put this under Development or NetworkServices ?
May 06 05:33:20 <dyoung> sounds like dev to me.
May 06 05:36:02 <jacques> hmmmmmmm
May 06 05:36:15 <jacques> I do wonder tho how fast qemu would run DOS on the slug
May 06 05:36:28 <jacques> never mind why :-D
May 06 05:39:48 <dyoung> so you can run ProCommPlus on it?
May 06 05:41:32 <jacques> I would do it just to say I did, tho that sort of thing can sometimes be leveraged into publicity for a project on /. (dunno if we want/need that tho)
May 06 05:41:50 <jacques> it's just pointless enough to get accepted
May 06 05:42:12 <dyoung> true.
May 06 06:03:08 <ecopperbech> msg Nickserv help
May 06 06:05:18 <copperbeech-work> jp30-work: ping
May 06 06:23:20 <[g2]> jacques and dyoung fyi http://www.nslu2-linux.org/wiki/HowTo/MountNFSSwapPartition
May 06 06:25:33 <jacques> thanks [g2] 
May 06 06:26:47 <[g2]> np
May 06 06:27:00 <[g2]> it *should* work
May 06 06:37:32 <NAiL> rwhitby: Soldering the router you're sitting on, or what? ;)
May 06 06:55:24 <jacques> here's something mildly interesting (which may be obvious, I dunno since I'm getting rather tired)
May 06 06:55:51 <jacques> I can run unslung binaries in qemu using the openslug libc
May 06 06:56:12 <jacques> (whereas unslung binaries do *not* run with the standard unslung libc)
May 06 07:06:06 <NAiL> ah
May 06 07:06:09 <NAiL> /dev/mtdblock4            6528      3772      2756  58% /
May 06 07:06:25 <NAiL> 2756kb free :)
May 06 07:06:52 <VoodooZ_Work> NAiL: Is this a result of removing switchbox?
May 06 07:07:26 <NAiL> VoodooZ_Work: Removing sqwitchbox only nets 128kb, afaik. I'ver built with uClibc
May 06 07:07:49 <[g2]> NAiL, did you build with uClibc ?
May 06 07:07:59 <NAiL> [g2]: yeah
May 06 07:08:05 <VoodooZ_Work> I see. I've been using glibc for months as I needed threads but I wonder if uclibc is faster?
May 06 07:08:07 * NAiL goes off to close bug #90
May 06 07:08:11 <[g2]> ls -l /boot
May 06 07:08:30 <NAiL> lrwxrwxrwx    1 root     root           21 Jan  1  1970 zImage -> /boot/zImage-2.6.11.2
May 06 07:08:33 <NAiL> -rw-r--r--    1 root     root       955624 Jan  1  1970 zImage-2.6.11.2
May 06 07:08:37 <[g2]> rm /boot/*
May 06 07:08:41 <[g2]> df
May 06 07:08:50 <NAiL> uh... Is that a bright idea? :)
May 06 07:08:55 <[g2]> sure
May 06 07:09:06 <NAiL> /dev/mtdblock4            6528      2828      3700  43% /
May 06 07:09:23 <[g2]> there you go!
May 06 07:09:27 <NAiL> Now *that* is free space ;)
May 06 07:09:36 <[g2]> it must be uClibc
May 06 07:09:48 <NAiL> yeah
May 06 07:09:49 <NAiL> with -Os
May 06 07:09:49 <[g2]> /dev/mtdblock4            6528      5116      1412  78% /
May 06 07:09:58 <[g2]> Ah...
May 06 07:10:29 <[g2]> I've still got the kernel in there to so I'll gain the 1MB back if I delete taht
May 06 07:12:06 * rwhitby freed up on 240V socket, and took the oportunity to upgrade the firmware in the wl500gx
May 06 07:12:11 <rwhitby> s/on/one
May 06 07:13:28 * [g2] is missing the socket part
May 06 07:13:35 <rwhitby> [g2]: I was soldering to use one wallwart for multiple devices
May 06 07:13:46 <[g2]> ah...
May 06 07:13:59 <[g2]> so the device was down anyway so the upgrade
May 06 07:14:17 <[g2]> I should do the same with my NSLU2s
May 06 07:14:26 <rwhitby> So AccessSlug, and it's 4 port usb 2.0 hub are now running from a single 2.5A wallwart, and the wl500gx and it's 7-port USB 2.0 hare running from the other 2.5A wallwart.
May 06 07:14:32 <[g2]> slug hydra cable :)
May 06 07:15:11 <[g2]> you forgot to mention the "on solar" part :)
May 06 07:15:29 <rwhitby> it took two tries.  The normal nslu2 power supply (2.0A) was not enough to run the wl500gx and the laptop disk attached to it's hub.
May 06 07:16:43 <[g2]> congrats
May 06 07:16:59 <NAiL> Yasy, the slug still boots ;)
May 06 07:17:10 <rwhitby> but it is enough to run accessslug and it's 4-port hub.  The bluetooth dongle seems to like a bit of power
May 06 07:17:27 <[g2]> nod, most RF devices do
May 06 07:18:24 <rwhitby> so the slug couldn't give the hub enough power for it, but tapping off the power supply before the slug did the trick.  go figure.
May 06 07:21:02 <rwhitby> will need to reset the router a couple of times to change settings ...
May 06 07:21:06 <NAiL> rwhitby: I do have some concerns regarding my current slug-mods though... How much power will an internal hub, usb->lpt converter and the display pull? 
May 06 07:21:14 <rwhitby> dunno
May 06 07:21:57 * NAiL isn't starting EE until this autumn, so I'm not all familiar with this yet ;)
May 06 07:26:22 <VoodooZ_Work> How would I force the loading of my ov511 camera driver on bootup as hotplug-ng doesn't handle coldplug events correctly yet.
May 06 07:50:28 <rwhitby-asleep> night all
May 06 07:51:54 <NAiL> nite
May 06 08:48:57 <VoodooZ_Work> Is the ARM's cache currently used in Openslug/Unslung? Wouldn't it helped with heavy sequential ram accesses?
May 06 08:50:08 <[g2]> VoodooZ_Work, it's used from the uncompress of the kernel on in Openslug
May 06 08:50:25 <[g2]> or are you talking about the 2K mini-cache
May 06 08:51:27 <VoodooZ_Work> so there's two cache? 
May 06 08:51:40 <VoodooZ_Work> L1/L2? or somet other linux specific one?
May 06 08:52:04 <[g2]> well there's an I-cache/D-cache and a 2K minicache specific to the IXP iirc
May 06 08:52:22 <VoodooZ_Work> Because I get major penalty for reading testing each pixel in an image and I remember reading the mini-cache was not enabled by default
May 06 08:52:36 <VoodooZ_Work> ok. Are we using both?
May 06 08:54:50 <VoodooZ_Work> anyways, lunch time. see you peeps later.
May 06 08:58:52 <NAiL> g2: ml+wiki done
May 06 08:59:00 <[g2]> thx
May 06 09:08:56 <NAiL> [g2]: What was the status of NTP btw?
May 06 09:34:40 <[g2]> NAiL, looking pretty good
May 06 09:34:55 <[g2]> jbowler was hot on the trail last night
May 06 09:35:19 <[g2]> there's a 4.2.0 version in oe that builds and runs
May 06 09:35:33 <[g2]> you need to scp over the ntpdate function
May 06 09:35:56 <[g2]> and there may be 1 issue with the time adjustments being negative
May 06 09:36:11 <[g2]> jbowler was all over it like white on rice
May 06 09:37:53 <[g2]> dyoung, was looking into the RTC initial valule also
May 06 09:43:24 <NAiL> [g2]: Ok. I'll try a build with glibc later tonight, and see how it pans out.
May 06 09:43:32 <[g2]> I'm running it now
May 06 09:43:40 <NAiL> eh... build with uClibc
May 06 09:44:17 <[g2]> ok, so you'll try ntp later built on the uClibc tool chain
May 06 09:44:58 <NAiL> yeah, right now my build box is busy doing glibc packages for fun. I'll go back to building samba and ntp using uclibc later 
May 06 10:01:07 <VoodooZ_Work> [g2]: Are we presently using the mini-cache? 
May 06 10:16:42 <[g2]> VoodooZ_Work, I think it may get used automatically
May 06 10:16:54 <[g2]> I just checked into it a little bit not too much
May 06 10:17:12 <[g2]> I was thinking for either packet throughput or usb throughput it would rock
May 06 10:21:17 <NAiL> [g2]: Uh.. how do I get the 4.2.0 version? Seems it likes 4.1.2 by default
May 06 10:22:50 <NAiL> Actually, I don't have a .bb for 4.2.0 at all
May 06 10:23:02 <VoodooZ_Work> yeah, definitely.
May 06 10:24:01 <VoodooZ_Work> Same for computer vision algorithms. very sequential.
May 06 10:24:41 <VoodooZ_Work> I was reading the TheSlugSpeedProblem wiki page and it makes you wonder if everything is setup right.
May 06 10:25:10 <VoodooZ_Work> I wonder if it would be possible to simply "overclock" the thing. I guess it depends if other functions share the crystal/clock
May 06 10:26:05 <VoodooZ_Work> Maybe I should ask kas11 about that stuff. She might know.
May 06 10:30:08 <macsat_bbl> I was wondering....what are the hardware specs (cpu mzh, ram amount and so on) of the NSLU2 ?
May 06 10:30:41 <NAiL> 32mb ram. CPU speed is either 133 or 266(?)
May 06 10:30:51 <macsat_bbl> ok
May 06 10:30:51 <NAiL> and 8mb flash, I think?
May 06 10:31:05 * NAiL *should* know this by now :P
May 06 10:31:06 <macsat_bbl> sounds reasonable with 8mb flash
May 06 10:31:09 <macsat_bbl> hehe
May 06 10:31:19 <macsat_bbl> I dont have one, so it is ok that I dont know ;-)
May 06 10:32:48 <NAiL> [g2]: Do you know if ntp 4.1.2 will work? 
May 06 10:34:25 <NAiL> (or are there necessary changes in 4.2.0?)
May 06 10:38:05 <[g2]> NAiL, dunno
May 06 10:43:19 <Jacmet> NAiL: yes, 8MB NOR flash
May 06 10:44:14 <Jacmet> (intel 28F640)
May 06 10:52:05 <CIA-1> 03caplink811 * 10unslung/Makefile: added package clamav to section native_ready_for_testing
May 06 10:53:11 <[g2]> Jacmet, are you a hw type ?
May 06 10:53:55 <NAiL> ntp+ntpdate takes 528kb
May 06 10:54:31 <[g2]> I'll bet you're happy you are building with uClibc :)
May 06 10:54:51 <NAiL> oh yes
May 06 10:55:05 <NAiL> But, for configuration. Did you guys put up a howto or anythiong?
May 06 10:55:37 <[g2]> really you only need to edit /etc/ntp.conf and add a distfile and three servers
May 06 10:55:52 <[g2]> it'd be good to write that up though
May 06 10:55:57 <Jacmet> [g2]: well, not really - 'm writing embedded sw for a living
May 06 10:56:18 <[g2]> Jacmet, cool me too :) what kind ?
May 06 10:57:15 <Jacmet> [g2]: various display technology - primarily flat screens for military use
May 06 10:57:38 <[g2]> sounds kinda fun
May 06 10:57:54 <Jacmet> [g2]: http://www.barco.com/defense/
May 06 10:58:01 <Jacmet> [g2]: and you?
May 06 10:58:08 <[g2]> that name sounds familiar :)
May 06 10:59:00 <[g2]> I did D&D in the 80's now I'm doing Open Source and looking to start a hw company
May 06 10:59:59 <Jacmet> [g2]: it can be ;) - military stuff has pretty crazy environmental requirements
May 06 11:00:15 <Jacmet> [g2]: ok
May 06 11:00:34 <[g2]> nod. 
May 06 11:04:26 <Jacmet> what magic is needed to compile the linksys kernel from nslu2-2.3R25.zip ?
May 06 12:05:37 <jp30-work> caplink811-away: ping?
May 06 12:05:53 <caplink811-away> jp30-work: pong
May 06 12:07:53 <jp30-work> i'm just testing clamav. i don't think that modifying /etc/passwd on the build system is a good way to deal with the problem. also i think you should have discussed the idea here or on the list before trying it
May 06 12:08:15 <jp30-work> ...is there a way to get clamav to build without a clamav user build time?
May 06 12:09:11 <caplink811> yeah, that could be, but, if i understand the docs right, clamav couldn't run as daemon if compile without the use
May 06 12:09:14 <[g2]> jp30-work, actually that's one of the big merits of OE in my book. you run everything as a normal user
May 06 12:09:16 <caplink811> s/use/user
May 06 12:09:44 <[g2]> no sudo's or anything
May 06 12:10:01 <jp30-work> [g2], same with unslung, which is why it didn't work
May 06 12:10:25 <[g2]> "which is a good thing(tm)" :)
May 06 12:10:43 <caplink811> jp30-work: may be possible to patch configure, wanna have a look on it
May 06 12:10:58 <jp30-work> caplink811, good idea, let's try
May 06 12:11:04 <jp30-work> [g2], agreed
May 06 12:12:05 <jp30-work> caplink811, perhaps it's possible to fool clamav into using a different user build time, then patch it back to "clamav" afterwards
May 06 12:12:58 <caplink811> 1st i tried cross on openldap, 2nd i have a look on clamav
May 06 14:11:45 <Christopher> anyone else having problems with the new version of apache and phpmyadmin?
May 06 14:12:28 <Christopher> A lot of pages in phpmyadmin only half load since I upgraded to the newer version of apache
May 06 14:12:36 <CIA-1> 03caplink811 * 10unslung/make/clamav.mk: builds now without a special user
May 06 14:13:50 <jf-away> jp30-work: ping
May 06 14:20:03 <Christopher> reducing the number of threads seems to have helped some
May 06 14:20:59 <jp30-work> jf-away, pong
May 06 14:22:22 <jp30-work> Christopher, do your problems date from apache 2.0.54 or from the transition to the worker MPM?
May 06 14:23:06 <Christopher> transition to worker mpm
May 06 14:23:37 <Christopher> apache seems to simply forget that it is sending the page
May 06 14:23:39 <jf-away> jp30-work: hi josh ! I was wondering if you would know what was the problem which make syslog listed in NEED_FOR_FIX wiley section before I saw it was also listed in the WL500G_PACKAGES
May 06 14:24:01 <jf-away> s/syslog/syslog-ng
May 06 14:24:35 <Christopher> oh i c it says that the php exceeded maximum execution time of 30s 
May 06 14:24:49 <Christopher> I'll have to see if I can change that in the conf somewhere
May 06 14:24:55 <jp30-work> Christopher, that sounds nasty. you weren't able to eliminate the problem by reducing # of threads, just reduce it?
May 06 14:25:38 <jp30-work> Christopher, ok that sounds like the problem
May 06 14:26:23 <jp30-work> let me know what fixes you make to your configs and i'll incorporate them into the apache / php packages
May 06 14:26:55 <Christopher> ok
May 06 14:27:13 <jf-away> btw, a few things : I remove mod-fastcgi.mk in WL500G_PACKAGES_READY_FOR_TESTING as apache is not (yet) listed in WL500G_PACKAGES, I promote eggdrop (should you have a look at the user creation stage) and last, do you know who is Inge Arnesen as its syslog-ng package is now in 1.6.7 and 1.6.6 is not downloadable anymore
May 06 14:27:34 <jp30-work> Christopher, one last qn, are you using eaccelerator?
May 06 14:27:44 <Christopher> yes
May 06 14:27:46 <Christopher> i think
May 06 14:27:50 <Christopher> I installed it
May 06 14:28:04 <jp30-work> jf-away, Inge is bob_tm
May 06 14:28:12 <Christopher> but I don't know a whole lot about it, so I don't know if it is doing anything
May 06 14:29:40 <jaro> does anyone have experience with php-thttpd?
May 06 14:29:59 <jp30-work> Christopher, eaccelerator caches php bytecode between script invocations so the scripts don't have to be recompiled. it is memory hungry, though. if it is installed, and you have restarted apache since, it should be doing its thing.
May 06 14:30:03 <jaro> I would like to know, wheter all dependencies are "must".
May 06 14:30:09 <jf-away> jaro : a few. 
May 06 14:30:14 <Christopher> ok
May 06 14:30:37 <jf-away> jaro : php-thttpd depends on php which depends on a lot of package
May 06 14:32:03 <jaro> jf-away : Yes. I see. But I would like to know, wheter these dependencies can be ignored in the special cases where those packages won't probably be used.
May 06 14:32:36 <jaro> jf-away: For example gdbm libdb libxml, libxslt
May 06 14:33:09 <jf-away> jaro: I think you may ignore them so. be carefull with libxml as it's use is wide in php
May 06 14:33:51 <jp30-work> jaro, jf-away, php and php-thttpd won't run at all without libxml
May 06 14:34:17 <jp30-work> same goes for libdb
May 06 14:34:26 <jf-away> as far as I remember, libxml is not a shared extension so that should be ok
May 06 14:35:47 <jp30-work> right, it's not a shared extension, that's why you can't get by without it
May 06 14:36:17 <jp30-work> ldd /opt/bin/php
May 06 14:36:17 <jp30-work>         libdl.so.2 => /lib/libdl.so.2 (0x4001d000)
May 06 14:36:17 <jp30-work>         libcrypt.so.1 => /lib/libcrypt.so.1 (0x40027000)
May 06 14:36:17 <jp30-work>         libdb-4.2.so => /opt/lib/libdb-4.2.so (0x4005b000)
May 06 14:36:17 <jp30-work>         libm.so.6 => /lib/libm.so.6 (0x4011a000)
May 06 14:36:18 <jp30-work>         libxml2.so.2 => /opt/lib/libxml2.so.2 (0x40195000)
May 06 14:36:20 <jp30-work>         libz.so => /opt/lib/libz.so (0x402a4000)
May 06 14:36:28 <jp30-work>         libc.so.6 => /lib/libc.so.6 (0x402c6000)
May 06 14:36:30 <jp30-work>         libpthread.so.0 => /lib/libpthread.so.0 (0x403d7000)
May 06 14:36:32 <jp30-work>         /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
May 06 14:37:11 <jp30-work> i could remove the gdbm and xslt extensions to their own ipks, but i figured that since those libraries are quite small, it wouldn't be worth it
May 06 14:39:17 <Christopher> jp30-work: It seems to be working better I increased the max_execution_time in php.ini to 60 instead of 30. I also increased the max mem usage from 8m to 16m
May 06 14:39:45 <CIA-1> 03jeanfabrice * 10unslung/make/dokuwiki.mk: Change the copy option from -R to -a in the dokuwiki-ipk target
May 06 14:40:05 <Christopher> jp30-work: probably not necessary to increase max mem, think it is mostly the execution time that was causing the problem
May 06 14:40:47 <jp30-work> Christopher, thanks. i'll certainly change the exe time setting. can you tell me more about how the mem use setting is affecting your system?
May 06 14:41:15 <jp30-work> ...ie. what size is your apache process after a fair bit of use with your modified setting?
May 06 14:41:30 <Christopher> k
May 06 14:43:40 <Christopher> jp30-work: so far, seems to be peaking around 45% according to top
May 06 14:44:36 <jp30-work> ok, does any of apache appear to be swapped out?
May 06 14:45:03 <Christopher> jp30-work: how would i figure that out?
May 06 14:45:42 <jp30-work> compare the resident size and virtual size of apache
May 06 14:49:29 <Christopher> top on the slug doesn't seem to give virtual size, is there another way to figure this out?
May 06 14:50:36 <Christopher> the resident size seems to max around 12m
May 06 14:50:45 <Christopher> I can do some more tests later, off to class now
May 06 14:51:22 <jp30-work> Christopher, ok, thx for your feedback, i'll put the max_time change into the php ipk
May 06 14:51:32 <Christopher-bbl> ok, thanks
May 06 14:58:30 <copperbeech> jp30: ping
May 06 14:59:51 <jp30-work> copperbeech, pong
May 06 15:00:44 <copperbeech> php-stage is still compiling - mysql has been compiling for 18 hours - but it is compiling... 
May 06 15:01:42 <jp30-work> good to hear that they native compile :)
May 06 15:02:12 <copperbeech> well it aint finished yet and I had to put the --with-low-memory switch in the makefile otherwise it failed on yacc
May 06 15:02:58 <copperbeech> it has been good to see that many of the other packages have compiled successfully in staging.
May 06 15:04:07 <copperbeech> bit of a crazy test of the native compiler trying to build php :-)
May 06 15:04:57 <jp30-work> yes, mysql even more so. some of mysql is c++, and g++ is very memory intensive
May 06 15:06:47 <copperbeech> yes but it seemed like the worst one is yacc (even has a little warning in the build saying that if it fails then to try the low memory switch)
May 06 15:16:14 <CIA-1> 03jp30 * 10unslung/ (make/php.mk sources/php/php.ini): php: increase max_execution time; slug is slow
May 06 16:22:32 * jstueve is building his first openslug-image
May 06 16:27:28 <jstueve> question: I have unslug 3-18 installed... what should I do to 'upgrade' to an openslug configuration.. start over from scratch?  will I have to backup/restore the hard drive?
May 06 16:30:55 <ByronT-Away> OpenSlug lets you do what you want to do
May 06 16:31:06 <ByronT-Away> you want custom sized partitions... go for it
May 06 16:31:31 <ByronT-Away> you want to use the Linksys partition setup... do it that way... your choice
May 06 16:31:53 <jstueve> I guess my question is.. do I need to rebuild the system/structure of my current disk if a switch to an openslug image?
May 06 16:32:15 <ByronT-Away> you don't have to, is the quick answer
May 06 16:32:28 <jstueve> The unslung packages are incompatible?
May 06 16:32:43 <ByronT-Away> correct
May 06 16:32:43 <jstueve> I think I read that on the wiki...
May 06 16:33:11 <jstueve> so anything I'm using, I'll have to restage.. is the directory structure the same as an unslung disk?
May 06 16:34:04 <jstueve> and I must say, for being Away you are very helpful! :D
May 06 16:34:41 <NAiL> jstueve: With openslug, you end up with a rather bare system. But then again, you can do pretty much whatever you want with it ;)
May 06 16:36:14 <jstueve> I don't have a serial or a jtag... will I even be able to use it?
May 06 16:36:19 <NAiL> yes
May 06 16:36:28 <NAiL> I have neither (failed building my serial cable)
May 06 16:36:30 <jstueve> and the IP address? defaults?
May 06 16:36:37 <NAiL> 192.168.0.77
May 06 16:37:14 <jstueve> that can be reset in the image?  my subnet is x.x.15.x
May 06 16:37:21 <NAiL> sure
May 06 16:37:41 <NAiL> sureu can change it before it boots the first time...
May 06 16:37:50 <jstueve> how do I do that?
May 06 16:38:04 <ByronT-Away> about the -Away... about to head home...
May 06 16:38:08 <NAiL> uh... Something strange happened there..
May 06 16:38:15 <NAiL> not sure you can change it...
May 06 16:38:27 <dyoung-web> The default address is 192.168.1.77
May 06 16:38:29 <NAiL> jstueve: join #openslug
May 06 16:38:59 <jstueve> okay...
May 06 16:39:15 <NAiL> aah.. yes. It remembered the ip from earlier...
May 06 16:39:37 <NAiL> <-- flashed earlier today, and thought I'd find it at .1.77, but found it at .0.77 ;)
May 06 17:29:46 <jacques> damn, I just got an email saying Les Suites (main OLS hotel) is almost sold out. I didn't expect it this early
May 06 17:30:21 <ka6sox-office> ugh...when is this event? this weekend?
May 06 17:30:22 <ka6sox-office> or next/
May 06 17:32:23 <jacques> ka6sox-office, OLS is in july (!)
May 06 17:32:33 <ka6sox-office> wow.
May 06 17:32:36 <jacques> i know
May 06 17:32:38 <jacques> :-(
May 06 17:32:50 <ByronT-Away> what+
May 06 17:32:55 <ByronT-Away> darn numlock on a laptop
May 06 17:36:05 <jp30-work> morning, eno
May 06 17:48:00 <jbowler> [g2], dyoung, dyoung-web: It turns out that on my OpenSlug NSLU2 the 'correct' tick is 10101.01 - but I have a fear that other slugs may need different values...
May 06 17:48:13 <eno> hi jp30-work 
May 06 17:48:39 <[g2]> jbowler, I read you e-mail and was trying out stuff now
May 06 17:48:49 <eno> my slug is not behaving today, had to reboot after up 45days
May 06 17:49:31 <jbowler> [g2]: if you want ntpq you will need to install the ntp-bin package too (and that needs perl, probably a mistake adding that as an RDEPENDS...)
May 06 17:51:23 <[g2]> jbowler, it's running
May 06 17:52:19 <jbowler> eno: terrible, still, my main slug file server has only bee up 53 days so far...
May 06 17:53:11 <jp30-work> anyone here know of a good tool for visualising differences between binary files? (must run on gnu/linux)
May 06 17:58:51 <NAiL> jbowler: Drift on my slug is 140.260
May 06 18:00:23 <NAiL> jbowler: How'd you figure out the 'correct tick'?
May 06 18:02:21 <jbowler> NAiL: run tickadj, it outputs a number (T), calculation T*(1+drift/1000000).  This is the correct number.  Edit /etc/default/rcS, change TICKADJ to this number.  run /etc/init.d/ntpd force-reload
May 06 18:03:58 <jbowler> So if you haven't changed anything, and everything worked, tickadj should output 10101 and the 'right answer' is 10102.  ntp can deal with that amount of error (140PPM) no problem, anyway, it will change after a few more hours of running.
May 06 18:04:38 <dyoung-web> need to glue an oven around that crystal.
May 06 18:06:13 <jbowler> dyoung-web: to see how temp sensitive it is?
May 06 18:06:59 <dyoung-web> well that too.  But to maintain a temperature.  Mostly though I'm just kidding around.
May 06 18:07:31 <ByronT> you say that, but I get the feeling you were thinking of how to mount a heat gun somewhere
May 06 18:08:25 <jbowler> Standard practice is to run boards in an environmental chamber for some specified time.  Acorn used to fry ARMs for a few hours a piece.
May 06 18:09:54 <dyoung-web> I was just looking at my box of oven crystals too.
May 06 18:10:10 <jbowler> That number - 10101 - is exactly 100/99*10000.  LinkSys (or someone) programmed something like 9901 somewhere inside the kernel to get the wrong tick.  It seems like someone calculated the inverse of the right answer.
May 06 18:10:41 <dyoung-web> hahahahahaha
May 06 18:10:44 <dyoung-web> thats classic!
May 06 18:10:52 <jbowler> Still, that's on the basis of a sample of 2 (my dev slug and NAiL's number).
May 06 18:12:06 <jbowler> Oh, I've just managed to boot 2.6.11-mm4 OpenSlug, with Reiser4 support.
May 06 18:12:16 <jstueve> hm...
May 06 18:12:53 <jacques> mmmmm reiser4
May 06 18:13:45 <jstueve> that is HOT...
May 06 18:13:50 <jstueve> whoops wrong channel.. ;)
May 06 18:14:37 <jbowler> Haven't got reiser4tools onto OpenSlug yet, but they worked fine before (on 'jonslug', same kernel).
May 06 18:16:25 <jacques> jbowler, your work on getting the slugs to tell time is greatly appreciated :-)
May 06 18:19:11 <jbowler> jacques: just send me a driftfile ;-)  If I get say 100 we'll know both whether there are board mod variations and how temp stable it is.
May 06 18:19:40 <jacques> jbowler, OK I'll generate a couple (have two slugs)
May 06 18:37:16 <barrym> Have people been able to get the new adduser package to respect newly created user ID's passwords after a powerfail?
May 06 18:37:48 <barrym> I'd add a slugbug, but there's nothing in there for it yet.
May 06 18:38:10 <barrym> I can create IDs fine, but upon reboot the passwords no longer work.
May 06 18:38:33 <barrym> This is on Unslung 3.18, I should add.
May 06 18:39:32 <dyoung-web> what does df / tell you
May 06 18:40:17 <barrym> # df /
May 06 18:40:28 <barrym> Filesystem           1k-blocks      Used Available Use% Mounted on
May 06 18:40:41 <barrym> rootfs                    6400      5308      1092  83% /
May 06 18:40:50 <barrym> #
May 06 18:42:18 <barrym> I'm trying to use "login" from my root id to test.
May 06 18:43:16 <barrym> # adduser barry
May 06 18:43:24 <barrym> adduser: barry: login already in use
May 06 18:43:34 <barrym> # login
May 06 18:43:43 <barrym> LKG0FD5B0 login: barry
May 06 18:43:52 <barrym> Password:
May 06 18:44:01 <barrym> Login incorrect
May 06 18:44:27 <barrym> Yet if I delete the "barry" ID, recreate it, and attempt it again, it works.
May 06 18:45:15 <dyoung-web> I havnt looked at adduser yet.
May 06 18:45:27 <dyoung-web> jp30 ping
May 06 18:45:52 <dyoung-web> hes kinda busy with something right now, but I think he got that package working right.
May 06 18:46:26 <barrym> Any idea what his login method was?  I'm using SSH.
May 06 18:49:02 <jp30-work> barrym, dyoung, jeanfabrice is your man for adduser
May 06 18:49:28 <jp30-work> ...i haven't tried using it myself
May 06 18:49:43 <jacques> I suspect this has something to do with the multiple passwd files in unslung
May 06 18:49:55 <jacques> rather than adduser per se
May 06 18:49:58 <barrym> Yeah, that's what I was thinking.
May 06 18:50:14 <barrym> I can't get SSH to recognise the IDs either.
May 06 18:50:15 <jp30-work> yes, that seems likely. 
May 06 18:51:12 <dyoung-web> is there a barrym user in /etc/passwd?
May 06 18:51:24 <dyoung-web> and if you add barrym2; does /etc/passwd become a file rather than a link?
May 06 18:52:09 <barrym> There is a barry and a barrym in /etc/passwd....
May 06 18:52:29 <dyoung-web> did you add them after or before your last reboot?
May 06 18:53:03 <barrym> Before.
May 06 18:53:17 <barrym> And /etc/passwd is a file, not a symlink.
May 06 18:53:36 <barrym> Should /etc/passwd be a symlink?
May 06 18:54:17 <barrym> From /etc/passwd...
May 06 18:54:19 <barrym> barry:x:2003:2004:Linux User,,,:/home/barry:/bin/sh
May 06 18:54:34 <barrym> barrym2:jCGgIIl4JGdWM:2005:2005:Linux User,,,:/home/barrym2:/bin/sh
May 06 18:54:44 <barrym> So for some reason the password isn't stored...
May 06 18:55:14 <dyoung-web> do you have a disk [lugged in?
May 06 18:55:24 <barrym> Yep.
May 06 18:55:46 <dyoung-web> can you look in/share/hdd/conf/passwd too?
May 06 18:55:55 <dyoung-web> usually /etc/passwd is a symlink to that flie.
May 06 18:56:37 <barrym> Hmm...  OK....
May 06 18:56:40 <jacques> I may be misremembering, but isn't the passwd file copied during shutdown ?
May 06 18:57:02 <barrym> In this case, /share/hdd/conf/passwd's barrym2 doesn't have a password.
May 06 18:57:23 <dyoung-web> okay thats your problem then/.
May 06 18:57:35 <dyoung-web> whihc is strange.
May 06 18:57:37 <dyoung-web> but anyways.
May 06 18:58:02 <barrym> So delete the /etc/passwd and symlink to /share/hdd/conf/passwd?
May 06 18:58:02 <dyoung-web> the q/d way to fix this right now is to edit the passwd hash into /share/hdd/conf/passwd .
May 06 18:58:12 <dyoung-web> the symlink will be recreated the next time you reboot.
May 06 18:58:52 <barrym> OK, I'll give it a shot.
May 06 18:59:20 <dyoung-web> gl.
May 06 19:00:18 <barrym> BTW, I took a crack at making "hpijs" for CUPS.  As far as I can tell, it basically compiles... but I have no idea if it really works, since CUPS still doesn't send data to my HP printer.  Anyone know the CUPS expert?
May 06 19:01:15 <dyoung-web> bob_tm
May 06 19:01:26 <dyoung-web> I think made the push with cups.
May 06 19:01:52 <barrym> Being a total noob at this, I have no idea what to do with this package I've made.  I'll send the file to someone for CVS input if they think it's useful.
May 06 19:03:12 <barrym> Cool.  I'll see if I can dig up his ID then.
May 06 20:59:27 <rwhitby-away> adduser is really meant to be run in an rc script by packages :-)
May 06 21:05:03 <rwhitby-away> we run it every boot to make sure the user is there for a package.
May 06 21:07:29 <jbowler> There's useradd as well, the whole thing is half-baked.  adduser has been used as a UI to simplify admin and as a generic API to mask machine independence.  Yuck.
May 06 21:08:35 <jbowler> There's still something strange about the openslug passwd/group stuff - I'm not sure that nsswitch.conf is working correctly.
May 06 21:27:47 <rwhitby-away> jbowler: BTW, thanks for fixing up the OpenSlug linuxrc.  I wasn't able to test it here due to a lack of a spare LAN port ...
May 06 22:42:32 <jbowler> rwhitby-away: I only did a minimal fix, there are still problems in the NFS stuff.
May 06 22:44:45 <rwhitby-away> yeah, we need someone who runs NFS rootfs on OpenSlug to fix that ...
May 06 22:45:26 <jbowler> Well, I can do that.  I can NFS mount with no problem.
May 06 22:45:40 <rwhitby-away> cool, thx.
May 06 22:45:52 <jbowler> But I'd like the /linuxrc to be simpler and to handle non-ext3 roots
May 06 22:46:15 <jbowler> 'cause my reiser4 system would use reiser4 on /
May 06 22:46:37 <rwhitby-away> ok, for openslug we don't really need to specify the -t ext3, and can just let it auto-detect.
May 06 22:47:09 <jbowler> Yes, but it seems that some fs types don't reliably auto-detect unless the module has been insmoded
May 06 22:47:42 <rwhitby-away> how do you propose we handle that then?
May 06 22:47:49 <jbowler> So my idea was for turnup to select the correct set of commands to mount the root (since something must happen to select a non-flash root)
May 06 22:49:03 <jbowler> I'd simply make turnup select the correct /linuxrc (e.g. by ln /boot/boot.nfs /linuxrc, /boot/* being the appropriate scripts).  Seems easier to understand for most people.
May 06 22:50:45 <jbowler> Easy to hack too for special purposes.  So long as the fallback is to flash (not sure how to do that if there is a partially ok boot, but I have a couple of ideas.)
May 06 22:50:58 <rwhitby-away> would each of those scripts handle the multi-level fail-safe fall-through we have in the linuxrc now?
May 06 22:52:12 <jbowler> Right: if the rootfs is obviously bogus it falls back to flash.  It's only really one level isn't it?
May 06 22:53:49 <rwhitby-away> yeah, I guess it is only one level, but done multiple times.  e.g. if it can't mount, then it fails safe, and if it can find the init, then it fails safe, and if it can't chroot then it fails safe.  A single level in three different places.
May 06 22:54:02 <jbowler> The fallbacks are easy to find by searching for 'exec' - those are the only possibilities.  exec chroot, exec init or the last line - /bin/sh
May 06 22:54:20 <rwhitby-away> I guess there is no reason why each of the /boot/boot.* scripts cannot do the same.
May 06 22:54:44 <jbowler> How about touch /boot/started, then in rmlogin rm -f /boot/started.  If /boot/started exists do a flash boot.
May 06 22:55:18 <jbowler> It touches the flash once on every boot - that's what I wasn't sure about.  It guarantees a failed boot returns to flash.
May 06 22:56:16 <jbowler> You can get into flash by starting a boot then pulling the plug before the LED goes green ;-)
May 06 22:56:35 <rwhitby-away> ok, the question is whether we want to do this for the 1.x beta release which is being prepared now, or wait for a 2.x ?
May 06 22:56:57 <rwhitby-away> we have recently made big changes, so there is no reason why we couldn't do this now too
May 06 22:57:28 <jbowler> Depends how long it takes to make it happen...  Most of the work is in getting NFS boot to go I think.  Scripting is easy.
May 06 22:58:10 <jbowler> big changes... It's highly testable.  Everyone will notice if it doesn't work ;-)
May 06 22:58:29 <rwhitby-away> it also means abandoning the idea of some commonality of boot procedure between openslug and unslung.
May 06 22:58:58 <rwhitby-away> but I think that has already gone past the point of no return anyway
May 06 22:59:45 <jbowler> Yes, the /linuxrc's are necessarily in different places in the BitKeeper repo, and ifconfig ixp0/ifconfig eth0 is almost unmanageable.
May 06 22:59:54 <rwhitby-away> yeah
May 06 23:00:16 <jbowler> I had a whole set of shell scripting to read /proc/interfaces and guess which one to use.
May 06 23:00:54 <jbowler> Hum...  My 2.6.11-mm4 openslug just went and hung again.  Several hours this time.
May 06 23:01:12 <dyoung> agree.  We're WAY past the point of common boot protocol.
May 06 23:03:36 <jbowler> The boot proto really is very simple now - simply removing the init=/linuxrc from the command line gives a successful boot.  It's a pity that RedBoot doesn't seem to allow this ('cause the command line is compiled in).
May 06 23:04:29 <jbowler> So the boot is a pure 'SystemV' boot sequence - the only thing missing is the ability to get boot messages before the network is up (S40network).
May 06 23:04:32 <dyoung> yeah.
May 06 23:05:19 <dyoung> jbowler, BTW, thanks for taking care of fixing that.  It was on my todo list; but I never got around to it.  Been busy with Real Life stuffs.
May 06 23:06:00 <ByronT> what?!? what is "Real Life?"
May 06 23:06:29 <dyoung> its that thing youre supposed to sqeeze in between sleep and play.
May 06 23:06:31 * ByronT sits to listen to dyoung's answer (and tells us in engrish)
May 06 23:06:35 <jbowler> What would help is to work out how to get netconsole working with the ixp425_eth driver - it's just something to do with missing polling support.
May 06 23:07:05 <dyoung> I thought Siddy had that working more or less?
May 06 23:07:29 <dyoung> he nailed the alignment issues with it, and patched some stuff.
May 06 23:09:01 <dyoung> Ah there it is.  http://www.nslu2-linux.org/wiki/OpenSlug/EnableNetconsole
May 06 23:09:50 <dyoung> the alignment patch is already in the main build tree somewhere.
May 06 23:11:49 <jbowler> The intdriven.patch checked in on Mar05 ([g2]) has part of this - the disable of polling at the end, but not the add of the poll controller which is what is missing at present.
May 06 23:12:46 <jbowler> I guess I could try adding that as well, but I have an MM patch which needs to go in first 'cause one of the functions is now declared in netdriver.h (and the module has void* not struct sockaddr*)
May 06 23:15:13 <jbowler> CONFIG_NET_POLL_CONTROLLER is set in the defconfig, so everything is there except the patch.
May 06 23:24:44 <jbowler> Well, it (ixp425_eth) compiles fine with the patch, and installs fine, will I be able to log in after a reboot though...
May 06 23:25:30 <dyoung> Cool.
May 06 23:25:46 <dyoung> There was an issue with netconsole, but I dont recall what it was.
May 06 23:28:01 <jbowler> The ixp425_eth works fine, on 2.6.11-mm4, problem is I have a mix'n'match build and I only have netconsole 2.6.11.2...
**** ENDING LOGGING AT Fri May 06 23:59:56 2005