**** BEGIN LOGGING AT Tue Mar 01 23:59:57 2005 Mar 02 00:03:28 morning! Mar 02 00:03:42 morning Mar 02 00:03:46 BREAKING NEWS: 2.6.11 is out! Mar 02 00:03:50 * siddy-away parties Mar 02 00:04:58 now we just need some volunteer to port it for openslug... :) Mar 02 00:05:56 really? Mar 02 00:05:57 shit Mar 02 00:06:38 ah, nice, it is tagged indeed Mar 02 00:06:42 thats great news Mar 02 00:07:10 siddy-away, a port is immiment. ;-) Mar 02 00:15:28 cool Mar 02 00:16:23 nothing on kernel.org yet Mar 02 01:36:21 well, there you go, nothing like a new kernel to provide an excellent reason for delaying openslug beta release. Mar 02 01:36:36 * peteru-home ducks and sneaks away :) Mar 02 01:37:04 peteru-home, it may just solve some of the outstanding problems. Mar 02 01:38:10 1.5-beta binary has already been built Mar 02 01:38:33 (couple of days ago) Mar 02 01:40:42 I need to put that in my new slug. Mar 02 01:41:21 considering that we'll probably want to move openslug to .11 at the earliest, it may make sense to start the public beta with a .11 based build. maybe? Mar 02 01:42:07 bbi1h - playing taxi driver... Mar 02 01:42:15 there is a significant EHCI patch in .11 Mar 02 01:42:25 well...its in .10 too Mar 02 02:42:12 peteru-home, I spent te evening looking at that. Mar 02 02:42:28 I have a 2.6.10 bb that works. Mar 02 02:42:59 Unfortnately there was a fundemental change in the way the arm timers are setup and we're looking at some work to bring the nslu2 specific stuff to that standard.... Mar 02 02:45:39 anyways, whenever they decide to post up the source tar/bz ball I'll start looking at it. Mar 02 02:46:03 for 2.6.11 ? yeah I noticed only the tgz was tehre Mar 02 02:47:07 really? I didnt even see that... Mar 02 02:47:15 yep Mar 02 02:47:25 it's been there for almost 3 hours (at least) Mar 02 02:47:47 ah ftp://ftp.kernel.org/pub/linux/kernel/v2.6/linux-2.6.11.tar.bz2 Mar 02 02:47:57 now they have the .bz2 Mar 02 02:48:34 Ahh...... wgetting now. Mar 02 02:48:39 me too :-) Mar 02 02:48:57 Are you motivated to get the bb working? Mar 02 02:49:26 huh? Mar 02 02:49:36 the 2.6.11 bb. Mar 02 02:49:44 (the one that doesnt exist yet) Mar 02 02:50:00 I have the feeling this one is going to be painful Mar 02 02:50:07 It will be. Mar 02 02:50:15 and beyond my skills Mar 02 02:50:18 If youre motivated, I have a bb for 2.6.10 that more or less works. Mar 02 02:50:39 the painful part is the nslu2 specific stuff which will need to be ported/converted to the new standard. Mar 02 02:51:06 I can mail you the 2.6.10 patchfile if you are interested to use that as a baseline. (if youre motivated...) Mar 02 02:51:08 there were tons of fundamental changes since 2.6.10 for arm Mar 02 02:51:25 I'm not motivated - it would be like banging my head against a wall Mar 02 02:51:28 not to mention USB changes. Mar 02 02:51:40 I'm kinda motivated. Mar 02 02:51:46 the kernel gurus will need to be involved in this one IMHO Mar 02 02:52:25 the fundemental timer changes were making my head spin a few hours ago. Mar 02 02:52:36 Its more under control now... Mar 02 02:52:43 (but doesnt make it suck any less) Mar 02 02:54:25 I think 2.6.11 will be a great arm kernel, but far from trivial to get running on the nslu2 Mar 02 02:55:15 I wonder if the new DMA mmap feature will be of any use to us Mar 02 02:55:37 heck they added arm TLB and smp support Mar 02 02:56:06 rearranged most of the arm assemby files Mar 02 02:57:21 like I'm going to have an ARM SMP box? Mar 02 02:57:58 very soon Mar 02 02:58:12 they did it because multicore arm are going to be available soon Mar 02 02:58:47 I am thinking of waiting until the dual-core A64 comes out. Mar 02 02:59:49 Okay so far its just the same 2 files that need restructuring. Mar 02 03:01:46 ka6sox-away, I thought that was at least 6 months - did you hear something ? Mar 02 03:02:05 3Q2005 Mar 02 03:02:10 so yeah. Mar 02 03:12:06 those should be so sweet Mar 02 03:13:52 peteru-home: the discussion over the last 30 minutes is why the 1.5-beta binary doesn't get rebuit on 2.6.11 ... :-) Mar 02 03:14:09 rebuilt ... Mar 02 03:15:46 nite all Mar 02 03:20:12 ~botsnack Mar 02 03:20:13 perlguru-work: aw, gee Mar 02 03:21:35 something intresting happened last 12 hours? Mar 02 03:34:07 hello Mar 02 03:34:25 does openssl work properly for you guys? Mar 02 03:34:36 yes Mar 02 03:34:38 i think there's still some big-endian/little-endian bugs in there Mar 02 03:34:42 which version are you using? Mar 02 03:35:27 0.9.7a vanilla seems borked Mar 02 03:35:55 i can ssh between armv4b/armv5teb boxes, but not from armv4b to x86 or vice versa Mar 02 03:35:57 the one in the unslung feed Mar 02 03:35:58 for example Mar 02 03:36:18 right Mar 02 03:36:24 no problems with openssh Mar 02 03:38:16 Package: openssl Version: 0.9.7d-2 Mar 02 03:39:36 aha Mar 02 03:39:38 i'll check that out, thanks Mar 02 03:39:54 np Mar 02 03:44:45 one more question if you don't mind: does unslung ship gmp/libgmp ? Mar 02 03:44:59 hey lennert Mar 02 03:45:26 hey rwhitby Mar 02 03:45:39 i figured only the dutch ppl would be awake round this time Mar 02 03:45:51 no, we don't have a gmp package in unslung Mar 02 03:46:04 it's only 10:15pm here Mar 02 03:46:50 ok, gmp is borked too, i wondered how you'd have solved it Mar 02 03:47:10 i fixed the 'floating point numbers are always little-endian' assumption but the testsuite still bombs out Mar 02 03:50:59 maybe there is a gmp in OE Mar 02 03:52:57 thanks Mar 02 03:56:37 gmp-4.1.2 in oe Mar 02 03:57:04 4.1.4 doesn't pass its testsuite Mar 02 03:57:08 i'll check it out Mar 02 03:57:14 maybe it doesn't pass the testsuite either :) Mar 02 03:57:17 thanks Mar 02 03:58:49 attempting first bitbake of openslug-kernel-2.6.11 Mar 02 03:58:57 brave man Mar 02 03:59:02 that was quick Mar 02 03:59:19 I'm setting up openslug-packages.bb Mar 02 03:59:25 I was gonna go to sleep, but i couldnt pass it up. ;-) Mar 02 03:59:40 rwhitby, what's openslug-packages.bb ? Mar 02 04:00:04 a place to capture the list of packages that work, so that we can upload automatically to a feed Mar 02 04:01:05 yeah, they seem to have released 2.6.11 while i wasn't watching :) Mar 02 04:01:33 lennert, dont fret too much, it was around 4 hours ago. Mar 02 04:01:52 i was asleep 4 hours ago Mar 02 04:01:57 lennert, any thoughts on 2.6.11 for arm? like "it should be good" or "avoid it" ? Mar 02 04:02:10 jacques: it has a number of important bugfixes for big-endian Mar 02 04:02:23 that sounds good Mar 02 04:02:52 and critically for us, it has a pile of ehci fixes over 2.6.9 . Mar 02 04:05:41 jacques: the eepro100 driver doesn't work though, since they ported it over to a new API but didn't implement the API for arm Mar 02 04:05:48 jacques: apart from that, it's all good and nice Mar 02 04:06:52 luckily we don't have one of those onboard :-) Mar 02 04:07:37 i'd choose it over the ixp4xx's npes any day though :) Mar 02 04:09:41 heh Mar 02 04:09:55 me too Mar 02 04:10:55 on the crossgcc list, glenn west (accidentally?) said (re the ixp4xx npe stuff) "working on scheduling another intel beating session soon with the local intel managers" Mar 02 04:11:14 robert ranslam (the intel guy responsible) replies "That would be me" Mar 02 04:11:28 Heh. Mar 02 04:11:46 lol Mar 02 04:11:51 I like gelloglue tho, hes seems like a nice guy. Mar 02 04:14:09 it's probably just the intel party line and not necessarily his decision Mar 02 04:16:03 Hmm, it complains less with 2.6.11 than 2.6.10.... Mar 02 04:27:52 Okay. now I'm back to where I was i 2.6.10. Mar 02 04:28:13 grumbling about gettimeoffset. Mar 02 04:28:25 what grumbling? Mar 02 04:28:27 and thats the stuff that needs restructuring. Mar 02 04:29:32 the needed mods for ixp2000 were fairly minor Mar 02 04:29:35 >=2.6.10 apparently doesnt use that variable/structure the same way as earlier.... Mar 02 04:29:55 Its not *that* bad, I dnt think.... Mar 02 04:30:00 but I'm not a guru. Mar 02 04:30:06 and I should be sleeping. ;-) Mar 02 04:30:19 is it in nslu2-specific code or generic ixp4xx code? Mar 02 04:30:31 nslu2-specific. Mar 02 04:30:51 so I would need to convert it to that standard. Mar 02 04:31:12 .timer etc Mar 02 04:31:40 anyone seen kas11? :-) Mar 02 04:31:42 well Mar 02 04:31:52 instead of INITTIME(ixp4xx_init_time) you say .timer = &ixp4xx_timer, Mar 02 04:32:04 one-line diff Mar 02 04:32:10 or are you using your own sys_timer ? Mar 02 04:32:19 own timer... Mar 02 04:32:39 any reason you can't just use ixp4xx_timer? Mar 02 04:32:47 not sure yet. Mar 02 04:32:51 most in-tree platforms seem to use it Mar 02 04:32:55 I havnt fully expored yet. Mar 02 04:32:58 explored. Mar 02 04:33:55 all in-tree ixp4xx platforms use it Mar 02 04:34:04 in 2.6.11 Mar 02 04:34:13 hmm... maybe tommorowm whenI'm kind of awake I'll comment that bit out to use the standard one. Mar 02 04:34:18 ok Mar 02 04:34:21 should work Mar 02 04:34:21 now I'm totaly brain dead. Mar 02 04:34:24 hehe Mar 02 04:34:30 sleep well then Mar 02 04:34:41 ( its 0234 HST ) Mar 02 04:34:46 Thanks Lennert! Mar 02 04:34:50 :) ciao! Mar 02 04:34:55 yeah, good progress so far dyoung-zzzz Mar 02 04:35:11 anyone feeling adventurous? I have a workable patchfile.... Mar 02 04:35:19 i'll have a look? Mar 02 04:35:22 but i don't have the h/w Mar 02 04:35:39 i'd be interested though Mar 02 04:36:46 cool :-) Mar 02 04:38:03 Okay sent! Mar 02 04:38:25 many thanks :) Mar 02 04:38:43 if you have an epiphany in the next 8 hours, reply to that email and I'll do something about it. ;-) Mar 02 04:39:12 lennert, do you have access to the bk://nslu2-linux source tree? Mar 02 04:39:20 is there supposed to be a X1205 driver in there? Mar 02 04:39:30 i don't use bk for religious reasons, sorry. Mar 02 04:39:33 Yes, nslu2 has a x105 in it. Mar 02 04:39:40 Okay, what about CVS? Mar 02 04:39:56 that would be fine, but i don't presently have access anywhere Mar 02 04:39:59 if you can CVS, the nslu2 specific files are available from sf.net Mar 02 04:40:04 aha Mar 02 04:40:08 project i openn2. Mar 02 04:40:24 i'll check that out Mar 02 04:41:00 nitey. Mar 02 04:41:04 ciao Mar 02 04:41:17 `night dyoung-zzzz Mar 02 05:04:29 03rwhitby 07 * r1.2933 10openembedded/ (4 files in 3 dirs): Sorted out licensing for openslug-image (including adding Info-ZIP to licenses.conf), and created openslug-packages. Mar 02 05:04:33 03rwhitby 07 * r1.2932 10openembedded/BitKeeper/ (etc/logging_ok deleted/.del-logging_ok~80f38da876ecf111): Merge Mar 02 05:04:37 03rwhitby 07 * r1.2931 10openembedded/BitKeeper/deleted/ (.del-logging_ok~80f38da876ecf111 .del-puppy_1.8.bb): Mar 02 05:04:37 Merge bk://nslu2-linux@nslu2-linux.bkbits.net/openembedded Mar 02 05:04:37 into home.(none):/home/openslug/openembedded Mar 02 05:06:30 peteru-home: don't be alarmed - that deleted puppy is one I started, but then realised you'd done it. your copy is intact in the repo. Mar 02 05:07:12 night all Mar 02 05:07:14 hmm, getting real close now Mar 02 05:07:16 night rwhitby Mar 02 05:07:23 but damn I'm sleepy. Mar 02 05:07:24 dyoung, weren't you going to sleep? :) Mar 02 05:07:32 i'm just looking at the openn2 cvs Mar 02 05:07:36 dyoung-zzzz won't sleep until it's done ... Mar 02 05:07:45 hehe Mar 02 05:07:45 he is a *MACHINE* Mar 02 05:08:17 be careful with that though - the last couple of weeks i found myself going to sleep when my girlfriend gets up, and starting to be awake by the time she goes to bed Mar 02 05:09:24 damn, list is turning into same FAQs asked over and over Mar 02 05:20:46 openn2 cvs is slightly.. confusing Mar 02 05:24:39 worse, it reverts a number of upstream changes Mar 02 05:50:58 morning. Mar 02 05:51:09 morning Mar 02 05:51:21 any progress with your amd issue? Mar 02 05:51:44 i have an image, made ext3 a module Mar 02 05:52:30 good. I'm still not happy about that workaround though. Mar 02 05:52:48 I'd rather fix the real problem before I start moving my work to my opteron. Mar 02 05:53:19 something is making the kernel bigger: linking or compressing Mar 02 05:54:08 I wanted to compare my xeon and opteron builds but it fails in e2fprogs for no freaking reason again so I've been lazy. Mar 02 05:55:01 i cannot mount :) Mar 02 05:55:15 tried to mount my hdd yesterday Mar 02 05:56:39 mount: Mounting /dev/sda1 on /mnt/hdd failed: No such device or address Mar 02 05:56:55 is there an sda1 file in /dev Mar 02 05:57:20 i cant check now Mar 02 05:58:16 Are you using udev? Mar 02 05:58:36 tried /media/hdd instead? Mar 02 05:59:08 cause /mnt/hdd doesn't exist in openslug unless you specifically created it .... Mar 02 05:59:09 yes same problem, thats why i tested with /mnt/hdd (which i created by hand) Mar 02 05:59:27 oh. ok. Is it a standard 1.5-beta build? Mar 02 05:59:41 yes, but with ext3 as module Mar 02 05:59:46 (and jbd) Mar 02 05:59:53 tried -t ext2 ? Mar 02 06:00:10 yes Mar 02 06:00:11 and auto Mar 02 06:00:22 what happened with -t ext2? Mar 02 06:00:40 same error Mar 02 06:00:51 ok, so it's not the modules then ... Mar 02 06:01:10 no those are loaded (checked with lsmod) Mar 02 06:01:26 doesn't matter whether they are loaded or not if ext2 doesn't work either Mar 02 06:01:37 mount = busybox mount Mar 02 06:01:51 will try again in a few hours Mar 02 06:02:05 what's in /proc/partitions? Mar 02 06:02:20 dont know, (no access to slug atm) Mar 02 06:02:57 check that, and make sure /dev/sda1 exists and has the right major and minor Mar 02 06:03:25 (when you get access, of course) Mar 02 06:03:32 i'll do that. thnx Mar 02 06:04:14 also check /proc/filesystems ... Mar 02 06:05:12 ok, i'll poke around in /proc Mar 02 06:30:10 perlguru-work: I got my opteron build to reach openslug-image (it fails because of size still) Mar 02 06:30:25 I'm now waiting for my fresh zeon build to complete so I can compare stuff. Mar 02 06:32:57 VoodooZ_work: try making ext3 a module Mar 02 06:33:18 its about 60k Mar 02 06:34:15 ok. Mar 02 06:34:47 Still. I want to compare to stock builds so I can compare stuff. Mar 02 06:35:58 phew... i've made all necessary appointments for importing my soon-to-be car Mar 02 06:36:04 compare the zImage-openslug thingies Mar 02 06:36:32 giel: what car? Mar 02 06:36:37 smart roadster Mar 02 06:36:45 :P Mar 02 06:37:01 Damn that xeon build is slow! Another reason to fix all this so I can cut my build time in 3! Mar 02 06:38:28 question is: are the kernel .o files also different? Mar 02 06:38:45 (sizes) Mar 02 06:38:54 good idea! Keep them coming. Mar 02 06:39:23 if those are the same, it good be linker or zipper Mar 02 06:39:26 * VoodooZ_work taps foot as he waits for the xeon build to finish packaging the zillions of glibc packages.... Mar 02 06:39:33 good -> could Mar 02 06:39:51 yep. good thinking. Mar 02 06:41:06 morning [g2] Mar 02 06:41:11 hi [g2] Mar 02 06:41:17 <[g2]> VoodooZ_work, perlguru-work Hey Mar 02 06:41:30 Ready for battle? :) Mar 02 06:41:48 <[g2]> This is a community right ? :) Mar 02 06:41:54 hehehe Mar 02 06:42:18 <[g2]> Waz up ? Mar 02 06:42:20 Yeah, but with any alpha s/w development it's a constant battle. Mar 02 06:42:37 [g2]: openslug package feed is up Mar 02 06:42:46 Trying to explain the size differences under AMD64 Mar 02 06:42:47 yes! Mar 02 06:42:58 what does that mean? Mar 02 06:43:02 src openslug http://ipkg.nslu2-linux.org/feeds/openslug/1.5-beta Mar 02 06:43:03 feed me ;) Mar 02 06:43:15 so you can install package using a URL? Mar 02 06:43:18 not many additional packages yet Mar 02 06:43:26 openssh? Mar 02 06:43:28 openssl, miau, atftp, ... Mar 02 06:43:29 bash? Mar 02 06:43:32 perl? Mar 02 06:43:41 very nice. Mar 02 06:43:55 apache? Mar 02 06:44:14 check packages/meta/openslug-packages.bb for the list Mar 02 06:44:30 <[g2]> rwhitby-asleep, Cool on the packages feed! Mar 02 06:44:37 and if you make some work, then add them to the list (and don't forget to add the symlinks to the symlinks repo too) Mar 02 06:45:13 <[g2]> Ok Mar 02 06:45:13 ipkg upgrade is much more interesting when you have all the packages for your base image in the feed ... Mar 02 06:45:48 <[g2]> rwhitby-asleep, 2.6.11 came out last night, dyoung was working 2.6.10 Mar 02 06:45:51 a lot of glib and locale stuff Mar 02 06:46:07 perlguru-work: that's all the base image stuff Mar 02 06:46:19 we basically just rsync up the tmp/deploy/ipk directory Mar 02 06:46:27 <[g2]> I'll be taking a look at 2.6.11 (it's dl'ing now) and I'd like to move to 2.6.11 (or at least have it available) Mar 02 06:46:29 a, i see. Mar 02 06:46:47 what do i need after a fresh flash? Mar 02 06:47:17 just put the src line in /etc/ipkg.conf Mar 02 06:47:56 where is coreutils> Mar 02 06:48:25 still in your fingers - you haven't typed it in yet ... Mar 02 06:48:29 <[g2]> rwhitby-asleep, .ixp... are in there. Mar 02 06:48:56 hmm Mar 02 06:49:21 did you see the discussion I had with kergoth in #oe about that? Mar 02 06:49:29 <[g2]> nope Mar 02 06:49:58 <[g2]> when does rikers log roll ? Mar 02 06:50:06 dunno Mar 02 06:50:18 how long ago was it? I have a live log Mar 02 06:50:28 0:00 GMT Mar 02 06:50:42 I have a part of it. Mar 02 06:50:49 <[g2]> !logs Mar 02 06:50:51 <[g2]> ~logs Mar 02 06:50:52 apt/ibot/jbot/purl all log to http://ibot.rikers.org// where channelname is html encoded ie: %23debian | lines that start with a space are not shown | some channels have stats at http://ibot.rikers.org/stats/.html.gz Mar 02 06:51:15 I can send it to you [g2] Mar 02 06:52:03 <[g2]> VoodooZ_work, thx Mar 02 06:53:11 <[g2]> boy it's sweet to get to kernel.org before the /. ing Mar 02 06:55:37 how do you delete from the beginning to the cursor in Vi ? Mar 02 06:56:03 <[g2]> VoodooZ_work, as in lines ? Mar 02 06:56:08 yep Mar 02 06:56:11 <[g2]> 1,.d Mar 02 06:56:20 ok. Mar 02 06:56:54 I gues to delete to the end is .,$d right? Mar 02 06:57:00 <[g2]> nod. Mar 02 06:57:20 <[g2]> After I typed it I was thinking ..... 1,$d Mar 02 06:57:26 <[g2]> as a joke Mar 02 06:57:46 * [g2] was a little too slow Mar 02 06:58:34 hehehe Mar 02 06:58:39 anyways, it's sent. Mar 02 06:58:46 <[g2]> thx Mar 02 06:58:51 np. Mar 02 06:58:51 ixp* are removed Mar 02 06:59:02 and excluded from future uploads Mar 02 06:59:23 * perlguru-work going home Mar 02 07:02:22 this is weird - at some point Ctrl-D stopped working for me Mar 02 07:02:59 <[g2]> rwhitby-asleep, thx until the matte is resolved Mar 02 07:03:24 agreed Mar 02 07:03:44 and the gloss too Mar 02 07:04:13 <[g2]> ;) Mar 02 07:07:11 03rwhitby 07 * r1.2933.1.1 10openembedded/packages/meta/openslug-packages.bb: Added bash and coreutils to openslug-packages Mar 02 07:08:34 hi Mar 02 07:08:44 g'day Mar 02 07:10:15 i have a really strange (wiring)- Problem with my NSLU2 and an 2,5"-PSU-less-diskdrive Mar 02 07:11:07 the disk is powered over the usb-port and the nslu isnt- "strong" enough Mar 02 07:11:33 yep, I had the same problem. Solved it with a 7-port powered hub. Mar 02 07:11:40 (the disk plugs into two ports) Mar 02 07:11:52 and with a powered usb-hub the disk run, but the disk isnt known Mar 02 07:20:23 That AMD64 size issue is odd. I just did two fresh builds on both my opteron and xeon boxes Mar 02 07:20:55 and almost every ipk package is bigger on the opteron side. The vmlinux file is ~100K bigger too. Mar 02 07:21:09 So maybe the whole cross-compiler is whacked. Mar 02 07:22:25 Any ideas what else I could compare to further isolate the issue? Mar 02 07:23:06 do you have your compiler flags set correctly? Mar 02 07:23:12 not generating debug symbols etc Mar 02 07:23:18 cross-compile hello world on both ... Mar 02 07:23:21 Well, I didn't change them. It's all OE's choosing. Mar 02 07:23:42 good idea rwhitby-asleep Mar 02 07:25:36 ok: On Dual Xeon: 7437 Mar 2 10:24 a.out Mar 02 07:25:48 on quad opteron: 7371 Mar 2 10:25 a.out Mar 02 07:26:36 Still a pretty big difference for such a small prog. Mar 02 07:27:02 all those extra zeroes. Mar 02 07:27:39 zeros? Mar 02 07:28:44 all the 64 bit padding Mar 02 07:28:44 * rwhitby-asleep is just making it up as he goes along ... Mar 02 07:28:51 heheh Mar 02 07:29:13 yeah, but it's only ran on a 64 bit machine, it's not targetting one. Mar 02 07:32:04 LOL - turbocharged_5 Mar 02 07:32:26 "Just what are all those folders for?" :-) Mar 02 07:52:50 Hello Mar 02 07:53:06 hello Mar 02 07:57:08 is somebody else using openslug and samba? Mar 02 08:09:59 chrta, there probably is, but they just haven't replied... Mar 02 08:12:19 i want to know if anyone receives a kernel panic under heavy load Mar 02 08:17:31 I also see occasional kernel-panics under heavy loads om openslug. Mar 02 08:17:44 what does your stackdump say? Mar 02 08:18:39 i think this: Unable to handle kernel NULL pointer dereference at virtual address 00000000 Mar 02 08:18:48 process smbd Mar 02 08:19:46 its inside the function kfree Mar 02 08:19:59 hm. my kernel used to crash somewhere in the cpu-scheduler. the stack was always totally smashed, so I had no clue where it really was. Mar 02 08:20:51 hi Mar 02 08:20:57 <[g2]> chrta, siddy-away 2.6.11 is out for a couple hours now, we are working on moving OpenSlug to that kernel Mar 02 08:20:59 my hope is that 2.6.11 fixed this problem! ;) Mar 02 08:21:04 i'm not very good in reading dumps from software i didn't wrote... Mar 02 08:21:13 hi perlguru Mar 02 08:21:27 hi [g2]! yeah, I read that! Mar 02 08:21:29 great g2 Mar 02 08:21:53 cat /proc/partitions Mar 02 08:21:58 8 0 9999864 sda Mar 02 08:22:06 <[g2]> dyoung was working on the 2.6.10 port some last night Mar 02 08:22:16 <[g2]> then he let me know 2.6.11 was out Mar 02 08:22:21 mount: Mounting /dev/sda2 on /media/hdd failed: No such device or address Mar 02 08:22:24 * siddy-away waits for dyoung-zzzz to wake up and continue the kernel-porting... Mar 02 08:22:35 no, dyoung already started with 2.6.11 Mar 02 08:22:47 perlguru: I'm discussing the AMD64 size issue in #oe Mar 02 08:22:47 with help from lennert Mar 02 08:22:56 VoodooZ_work, ok Mar 02 08:23:11 <[g2]> siddy-away, that was after his work on .10 Mar 02 08:23:32 the cross compiler under AMD64 is out of whack and even the hellow world program produces bigger binaries. Mar 02 08:24:07 strange, it shouldn't, because target is same Mar 02 08:24:38 is it possible to compile samba using uclibc? my do_compile step fails Mar 02 08:24:39 <[g2]> timers changed in .10 -- I was going go get the latest OpenN2 stuff and probably pull the timer code in there so it just does GPIO Mar 02 08:25:17 <[g2]> chrta you'll need to rebuild everything on that TARGET_OS chain Mar 02 08:25:29 good luck with that, the latest versions of OpenN2 are quite different. Mar 02 08:25:56 That's why I only used rev. 0.7 in openslug-kernel-2.6.9 Mar 02 08:26:10 kas11 changes quite a few things in newer versions. Mar 02 08:26:10 <[g2]> All we need is GPIO support for the LEDs Mar 02 08:26:38 As long as I2C works I'm happy. Mar 02 08:27:28 <[g2]> I've started looking it, I'm not sure all the things that need to be done Mar 02 08:27:49 /proc/bus/usb is empty with hdd attached :/ Mar 02 08:29:33 * VoodooZ_work loves having the latest but hates dealing with the bugs :) Mar 02 08:29:47 not bugs -> features Mar 02 08:32:33 hehe Mar 02 08:34:51 the #oe guys seem to think that the cross-compiler under AMD64 is trying to build 64bits ARM binaries or as they have twice the word size. Mar 02 08:36:10 why does gcc 3.4.3 do that, gcc 3.3.4 did not have that problem Mar 02 08:36:16 3.3.5 Mar 02 08:36:41 really? Mar 02 08:36:47 That could be a hint. Mar 02 08:36:51 and why does my kernel work? Mar 02 08:37:10 or is my problem maybe related? Mar 02 08:37:48 It's a pain to figure out as I'm no expert. Mar 02 08:37:50 :( Mar 02 08:38:13 Even the experts in #oe have no clue so it scares me. :) Mar 02 08:41:59 the problem may be that a compiler is everything but deterministic... Mar 02 08:42:25 it depends quite much on the architecture it was built on. Mar 02 08:43:11 AMD64 in this case Mar 02 08:43:35 yeah. Mar 02 08:44:24 yeah... the root of the problem may be everywhere... perhaps it's the cross-compiler's error, or a problem with the compiler that compiled the cross-compiler or Mar 02 08:44:37 ouch. Mar 02 08:45:15 ever read "Reflections on Trusting Trust"? Mar 02 08:45:47 nope. Mar 02 08:45:51 it gives a great insight on the amount of trust you actually set in your compiler: http://cm.bell-labs.com/who/ken/trust.html Mar 02 08:49:06 That most everything else like my computer, my car or even my alarm clock! Mar 02 08:55:53 my samba build failes :( Mar 02 08:56:10 i already deleted the tmp and build/tmp folder Mar 02 08:56:12 VoodooZ_work, you can try the #gcc channel Mar 02 08:56:36 I guess. Mar 02 08:57:12 <[g2]> chrta, if samba doesn't build with uClibc and everything else does andersee and others in #uclibc will want to know about it Mar 02 08:57:16 on which server? freenode? Mar 02 08:57:26 <[g2]> chrta that's on freenode Mar 02 08:58:08 I mean the "official" gcc IRC channel? Mar 02 08:58:21 ok, i'll try it Mar 02 08:58:31 <[g2]> chrta you'll also want to check the uClibc mailing list too Mar 02 09:06:34 [g2], can you add more buzzer feedback during boot? (also when boot is done) Mar 02 09:07:10 like: beep-beep when it starts, and beep when done Mar 02 09:07:24 (i like beeps) Mar 02 09:08:59 <[g2]> perlguru, the buzzer uses the timer interrupts it hasn't be tested hardly at all Mar 02 09:09:25 <[g2]> I don't think risking the system for the feedback it provides is worth it right now Mar 02 09:09:26 [g2], it beeps now ehen the boot is started Mar 02 09:09:47 I thought the buzzer used delay loops? Mar 02 09:09:50 <[g2]> I think that a loop, but I'd need to double check Mar 02 09:09:51 i dont see the slug now, it is under my desk Mar 02 09:10:01 bbl Mar 02 09:10:57 bbl lunch time Mar 02 09:32:10 the samba problem is a conflict between ipc.h and posix_types.h. i fixed it on my local host. Mar 02 09:34:07 chrta: is that the __kernel_key_t failure? Mar 02 09:34:27 chrta: thats bitten a number of packages that include kernel headers when they shouldnt be. they're just lucky that they build for glibc Mar 02 09:34:30 heh Mar 02 09:35:22 yes, it's this error Mar 02 09:35:57 not sure what the best solution is. userspace packages should really avoid including kernel headers. Mar 02 09:36:13 andersee would argue for fixing the apps to include a local set of headers, no doubt Mar 02 09:36:15 i included posix_types in ipc now and removed one typedef in ipc Mar 02 09:37:14 btw, i dont know anything about the kernel structure... Mar 02 09:51:35 the build of cups fails, too :( Mar 02 09:53:35 oe build errors really belong in #oe, not here Mar 02 09:55:45 sorry Mar 02 09:56:03 not a problem, just figured i'd let you know Mar 02 10:20:54 hey kergoth Mar 02 10:21:57 hey Mar 02 10:22:24 did you see the discussion about the AMD64 based builds producing larger binaries in #oe 2 hours ago? Mar 02 10:22:57 RP, emte and cosmicpenguin were saying it looks like the cross-compiler on amd64 produces 64 bits ARM binaries or something. Mar 02 10:23:21 I've done comparison between my xeon and opteron boxes and almost everything is bigger. Mar 02 10:23:44 It's mostly the .text section so somebody suggested it might be the word-size. Mar 02 10:23:58 ah, thats no good Mar 02 10:24:00 anyways, I thought you might have ideas. Mar 02 10:24:03 hm Mar 02 10:24:13 ask me again when i've had something caffinated Mar 02 10:24:14 I can run some test if you have ideas. Mar 02 10:24:18 hehehe Mar 02 10:24:21 ETA? Mar 02 10:24:33 in 1 hour? Mar 02 10:24:47 thatd be fine. probably sooner, but i've got real work to do too Mar 02 10:24:54 ok. thanks mate. Mar 02 10:25:33 also explain why unslung works for me and openslug not Mar 02 10:25:44 ummm. another curve ball. Mar 02 10:25:46 only diff: gcc 3.3.5 - 3.4.3 Mar 02 10:25:57 yep. it seems to be related to 3.4.3 Mar 02 10:26:25 the #oe guys were also talking about 32/64 bit compatibily libs. Mar 02 10:26:41 than unslung should have same prob Mar 02 10:27:24 or the crosscompiler is build different for unslung and OE Mar 02 10:28:16 meanwhile i reflash with unslung Mar 02 10:28:58 perlguru: what are you talking about? Mar 02 10:29:11 he just told you, the problem is specific to 3.4.3 Mar 02 10:29:32 i have no problems with crosscompiled packages for unslung (not OE) Mar 02 10:29:55 unslung uses gcc 3.3.5 Mar 02 10:30:24 if unslung uses gcc 3.3.5, that isnt 3.4.3, now is it? Mar 02 10:30:29 well, i'm not absolutely certain it's specific to 3.4.3 but that's what perlguru mentionned. Mar 02 10:30:58 VoodooZ_Work: have you had a chance to pester any of the gcc hackers on irc yet? Mar 02 10:31:01 That it stopped working after the change to 3.4.3 under x86-64 Mar 02 10:31:18 on the gcc chanell you mean? Mar 02 10:32:08 yeah Mar 02 10:32:21 dunno how popular the one on freenode is, but iirc the #gcc channel on oftc is pretty popular Mar 02 10:32:22 ok. I'm in uclibc already so I'll check it out. Mar 02 10:32:35 perfect. I was about to ask that. Mar 02 10:33:18 oftc? x-chat doesn't list that one. Mar 02 10:34:26 ~oftc Mar 02 10:34:27 i heard oftc is the Open and Free Technology Community, a support/collaboration service. They have an IRC network: irc.oftc.net Mar 02 10:34:57 ~botsnack Mar 02 10:34:58 perlguru: aw, gee Mar 02 10:34:58 a bunch of channels moved there Mar 02 10:35:02 thanks Mar 02 10:35:05 when they got sick of lilo's begging for funds Mar 02 10:35:10 like #kernelnewbies Mar 02 10:35:13 I'll add it. Mar 02 10:46:29 #gcc Mar 02 11:07:03 * perlguru running unslung again Mar 02 11:08:57 bye Mar 02 12:49:26 03jp30 * 10unslung/make/espgs.mk: remove line from espgs.mk that would break other packages Mar 02 13:36:54 well, it built. Mar 02 13:37:07 gonna try build the rest of the image and boot it now. Mar 02 13:39:35 <[1]garpinc> jp30 which line? Mar 02 13:41:18 Christ. Mar 02 13:41:20 Its too big. Mar 02 13:45:44 dyoung-zzzz, that's what I feared... how big did it get? Mar 02 13:47:36 Well, as configured (ie: before deleting crap) its 1107176 Mar 02 13:47:58 so I just need a little bit to squeeze it in. Any suggestions on what to yank real quick and have it still work? Mar 02 13:48:19 do you need ext3 for booting? Mar 02 13:48:20 Hows about these host-to-host cables. Mar 02 13:48:46 Well, kind of. I dont *need* it to boot, but I need it to test this EHCI stuff. Mar 02 13:49:02 make it a module. Mar 02 13:49:36 Yeah, but then I'll need to diddle more files to actually get themodule in the rootfs. I'm going for instant gratification here. Mar 02 13:49:37 heh Mar 02 13:49:58 removing all those strange network-protocols (appletalk etc pp) from defconfig also helps a bit. Mar 02 13:50:26 but I fear it wont shrink it by 60k... :-\ Mar 02 13:50:38 Well, I yanked ext3 for now Mar 02 13:50:45 just to see if it'll boot. Mar 02 14:10:59 <[g2]-away> dyoung-zzzz, I'll be back in a bit, I started work on the 2.6.11 kernel today thx for your e-mail Mar 02 14:11:24 [g2]-away, I JUST made a image, flash ing now. Mar 02 14:12:21 [g2]-away, if you or someone can push a change to openslug-image to use a preferred kernel of 2.6.9 that would be good. Mar 02 14:12:40 so it wont automagically use the 2.6.new Mar 02 14:13:36 Bah, no joy. Mar 02 14:15:43 dyoung-zzzz, not working? Mar 02 14:16:21 it loads the kernel but fails to boot. Mar 02 14:16:32 done, booting. Then nothing. Mar 02 14:17:05 hm, not nice. :( Mar 02 14:17:13 I'm not that suprised. Mar 02 14:23:08 hm... my battery is empty... time to sleep! g'nite everybody! Mar 02 14:23:28 nite. Mar 02 14:47:12 <[g2]-away> dyoung, I gotta run soon, but what's happening on the boot ? Mar 02 14:47:36 <[g2]-away> did you remove the openslug-init call to leds ? Mar 02 14:47:49 no Mar 02 14:47:49 <[g2]-away> I was going to remove the timer stuff from the nslu2-io Mar 02 14:48:00 it says ..... Done. Booting. Mar 02 14:48:03 then nothing. Mar 02 14:48:46 <[g2]-away> isn't there like the first printk Mar 02 14:50:10 <[g2]-away> Uncompressing Linux..................................................................... done, booting the kernel. Mar 02 14:50:10 <[g2]-away> Linux version 2.6.9 (tom@gentoo2) (gcc version 3.4.3) #2 Mon Feb 28 21:01:43 EST 2005 Mar 02 14:50:29 No, it dont get that far. Mar 02 14:50:32 <[g2]-away> And you don't get the first printk which is the Linux Mar 02 14:50:34 <[g2]-away> ... Mar 02 14:50:36 nope Mar 02 14:50:45 unless its scrolled off the screen which I doubt. Mar 02 14:50:46 <[g2]-away> Linux version 2.6.11 .... Mar 02 14:51:03 so it jumped to the decompressed kernel image and died? Mar 02 14:51:13 there are so many possible causes of that, i wouldnt even know where to start debugging Mar 02 14:51:17 heh Mar 02 14:51:43 <[g2]-away> That happens PRETTY early last I checked Mar 02 14:52:04 Yep, it goes Mar 02 14:52:06 Uncompressing Linux.................................................................... done, booting the kernel. Mar 02 14:52:13 Then thats it. Nothing. Mar 02 14:52:51 <[g2]-away> Same kernel paramters right ? Mar 02 14:53:00 yes Mar 02 14:53:00 <[g2]-away> kernel cmdline paramters Mar 02 14:53:22 <[g2]-away> which patches do you have applied Mar 02 14:53:34 from the patchfile I sent you. Mar 02 14:53:44 that includes everything that was in 2.6.9 Mar 02 14:53:55 (that isnt already in there of course) Mar 02 14:54:25 <[g2]-away> did you change any of the nslu2-io stuff ? Mar 02 14:54:45 Yes, to make it compile, which is why I'm not all that suprised it doenst boot. Mar 02 14:55:02 <[g2]-away> Ok... I've got to run, I'll be back in several hours Mar 02 14:55:08 ok Mar 02 14:55:09 me too. Mar 02 14:55:12 <[g2]-away> THX for your help on this Mar 02 14:55:23 <[g2]-away> It's just like the old days :) Mar 02 16:58:34 [g2]-away: it compiles without the nslu2 specifics, I cant test it for another 5 hours or so though. Mar 02 17:02:14 [g2]-away: and a possible reason why I cant boot is the rootfs it made is enormous for some reason. Mar 02 18:17:29 dyoung-web, what's the latest on the 2.6.11 -> openslug? Mar 02 18:19:39 * dyoung-web mumbles random obscenities. Mar 02 18:20:24 oh, so your day was as *good* as mine... Mar 02 18:20:50 in other words, it was just this side of a nightmare Mar 02 23:19:19 * peteru-home is excited by this box: Netgear WGT634U Mar 02 23:19:34 any opinions on it? Mar 02 23:21:14 the bootloader in particular looks very sweet. Mar 02 23:27:20 dang.. why does the nslu2 udhcp not put search domains into resolv.conf.. Mar 02 23:27:54 if I have too many of them on the dhcp server, it wont even get online Mar 02 23:29:35 (also the fdisk doesnt work, known problem perhaps) **** ENDING LOGGING AT Wed Mar 02 23:59:56 2005