**** BEGIN LOGGING AT Thu Feb 15 02:59:57 2007 Feb 15 03:20:34 last post on the website was from 11 months ago. the mailing list doesn't seem too active lately. Feb 15 03:20:41 is angstrom actively developed? Feb 15 03:33:15 http://cia.navi.cx/stats/project/openembedded/ Feb 15 03:33:20 very actively Feb 15 03:40:45 :) Feb 15 03:44:12 I'm having a peculiar difficulty keeping a wlan connection stable, so I will be checking out and building Angstrom for my c3100 soon Feb 15 03:46:07 rwhitby: do you know anyone who would like to have hands-on access to the zaurus remote for cxx0 ? Feb 15 03:46:23 I can loan mine out for a few months no problem, if it will help encourage support Feb 15 03:48:13 jnc: dunno - ask koen|away Feb 15 03:48:34 koen|away: delightful ping :) Feb 15 05:40:57 actually, that anime is mediocre at best, but the manga is much better. Feb 15 05:41:05 (marex quit) Feb 15 13:48:18 <_law_> hm any ideas why udev cant mount /dev there is some wrong fs-type error on system boot... Feb 15 13:54:58 there is still the bug where GPE image on spitz does not create a /home/root directory Feb 15 13:55:11 it messes up the root term home directory Feb 15 13:55:26 other than that, yesterday's checkout is running and looking fabulous Feb 15 13:55:28 nice work Feb 15 13:56:11 rvalles: I am also a fan of Ah! Megami Sama Feb 15 13:56:26 not the manga, I don't have patience for that Feb 15 13:56:28 :/ Feb 15 13:58:33 <_law_> where and how will /dev be mountet? Feb 15 13:59:33 _law_: /etc/init.d/udev mounts a tmpfs over /dev Feb 15 14:02:49 hmm Feb 15 14:03:10 jnc: but the manga is so much better! Feb 15 14:03:13 do either of you use a D-link dcf660w wireless cf card? Feb 15 14:03:22 jnc: I have one Feb 15 14:03:33 with an older build from months ago, it sort of worked Feb 15 14:03:35 jnc: you know Urd doesn't like people who don't read it? I'd rather not upset Urd. Feb 15 14:03:39 now it won't even flash the green lights Feb 15 14:03:50 any idea? there's dmesg output saying it can't find the pri f/w Feb 15 14:04:01 so reinsert card? Feb 15 14:04:13 my card is broken Feb 15 14:04:16 tried that. tried rebooting with it in, tried rebooting with it out Feb 15 14:04:26 I think it might be the kernel honest Feb 15 14:04:35 or the firmware is not compatible Feb 15 14:05:15 my troubles with the card started when I moved to a newer build of Angstrom and a new kernel version some months ago Feb 15 14:05:59 <_law_> pH5: thanks on boot i get the following error... Starting the hotplug events dispatcher udevd Feb 15 14:06:17 I had this working to authenticate with my WAP just yesterday, with a build of angstrom from 2006-11-17 Feb 15 14:06:34 it would frequently cut out on me though Feb 15 14:06:39 <_law_> mount: wrong fs typw, bade option, bad superblock on /dev Feb 15 14:08:57 _law_, I'd recommend you to compare you defconfig with "known good" Feb 15 14:09:36 hrw: dmesg output at http://rafb.net/p/8TNyN087.html Feb 15 14:09:55 <_law_> psokolovsky: ok but perhaps the other working image doesnt use udev? Feb 15 14:10:32 _law_, no, all images use udev, and all work. and issues like this are old ;-I Feb 15 14:10:45 <_law_> psokolovsky: i use a fat32 formated sdcard could this be the problem? Feb 15 14:11:18 jnc: try hostap-utils which provide flash tool Feb 15 14:11:28 _law_, for what? Feb 15 14:11:29 hrw: I will do that, thanks Feb 15 14:11:43 hrw: just re-flash the firmware? any suggested version? Feb 15 14:12:35 <_law_> psokolovsky: udev cant mount my /dev Feb 15 14:13:06 <_law_> psokolovsky: no couldnt its /dev/ram Feb 15 14:13:35 _law_, you use fat32 for what? Feb 15 14:13:57 cat /proc/filesystems | grep tmpfs # missing tmpfs support in the kernel? Feb 15 14:14:00 <_law_> psokolovsky: my sdcard where i boot the angstrom image Feb 15 14:14:09 _law_: can you check what exact mount command fails? Feb 15 14:15:07 _law_, you have linux rootfs on fat32 filesystem? that won't work Feb 15 14:15:36 <_law_> pH5: Starting the hotplug events dispatcher udev is the last outbput Feb 15 14:15:41 <_law_> output Feb 15 14:16:21 <_law_> pH5: line 171 from /etc/init.d/udev Feb 15 14:17:55 jnc: 1.7.4 STA, 1.1.1 PRI Feb 15 14:19:16 <_law_> psokolovsky: set cmdline "ip=10.0.0.2:10.0.0.1:10.0.0.1:255.255.255.0:axim:usb0:none console=ttyS0,115200n8 console=tty0 ramdisk_size=16384 rdinit=/sbin/init psplash=false" Feb 15 14:19:34 <_law_> sould i also set root=/dev/ram ? Feb 15 14:20:09 hrw: oh, and it would save me a trip, is setting that special option in the kernel module source code necessary with the recent kernels from angstrom? last time I did the firmware change it required me to change a line of code and compile a special module, I should do this now? Feb 15 14:20:13 jnc: in other words: latest PRI, 1.7.4 STA Feb 15 14:20:27 jnc: probably same Feb 15 14:20:37 <_law_> psokolovsky: or /dev/ram0 Feb 15 14:20:49 _law_, so, you don't have rootfs on fat32, you have it in initramfs Feb 15 14:21:35 <_law_> psokolovsky: right but zImage and rootfs.cpio.gz is on fat32 Feb 15 14:21:36 _law_, so well, oops, I'm sorry if that's me who actually caused you troubles. I booted opie only in such mode (and bootstrap images of course). Feb 15 14:21:49 _law_, that doesn't matter for linux, of course Feb 15 14:22:35 _law_, I heard something about initrmafs being ro. try adding "rw" on command line, and ssh into it and check how it is mounted Feb 15 14:23:08 _law_, can you ssh into device at all? Feb 15 14:23:29 <_law_> psokolovsky: not yet no /dev/random /dev/urandom Feb 15 14:23:39 <_law_> psokolovsky: also telnet doesnt work Feb 15 14:23:58 <_law_> but pinging the device works Feb 15 14:24:12 <_law_> an graphical output also Feb 15 14:24:20 _law_, angstrom images come with sshd (dropbear) Feb 15 14:24:45 <_law_> psokolovsky: but dropbear doest start because there is no /dev/random Feb 15 14:24:57 _law_, that's why I suggest you looking at good defconfig. like h4000_defconfig Feb 15 14:25:06 _law_: how about adding /dev/random to initramfs? Feb 15 14:25:55 <_law_> pH5: how Feb 15 14:26:11 Is it the tmpfs mount that fails, or the bind mount for /dev/.static? Feb 15 14:26:12 <_law_> psokolovsky: how should i set initramfs to rw? Feb 15 14:26:22 _law_: OE dropbear use /dev/urandom Feb 15 14:26:41 _law_: I don't know. How do you generate your initramfs? Feb 15 14:26:51 _law_: add 'rw' to the kernel command line. Feb 15 14:27:31 _law_, hey, indeed, how did you generate it? ;-) Feb 15 14:29:34 <_law_> psokolovsky: set initrd "Angstrom-bootstrap-image-test-20070209-aximx50.rootfs.cpio.gz" Feb 15 14:29:47 <_law_> psokolovsky: rdinit=/sbin/ini Feb 15 14:29:50 <_law_> t Feb 15 14:30:09 _law_, ok, good Feb 15 14:31:16 _law_, so, I confirm that defaulkt static /dev in agstrom lacks random/urandom Feb 15 14:31:56 <_law_> psokolovsky: right Feb 15 14:33:38 _law_, and, quick look in kernel source shows that /dev/random and /dev/urandom are builtin, present in any kernel Feb 15 14:33:53 obj-y += mem.o random.o tty_io.o n_tty.o tty_ioctl.o Feb 15 14:34:07 _law_, so, you need to figure what's wrong with udev Feb 15 14:34:59 _law_, otherwise, you can indeed unpack cpio.gz, mknod random/urandom, repack, book, ssh and investigate in comfort on device ;-) Feb 15 14:36:58 pH5, Xw100 hangs for me on quit/suspend/whatever. can you confirm? Feb 15 14:41:36 <_law_> i which .bb file /dev devices will be build? Feb 15 14:42:02 <_law_> so i can add random and urandom for testing Feb 15 14:42:30 psokolovsky: I don't have the hx with me right now. I could try this weekend. Feb 15 14:42:36 _law_, don't do that. find why udevd doesn't start for you Feb 15 14:42:49 pH5, ok, thanks Feb 15 14:43:07 psokolovsky: don't you think (u)random should be in static /dev anyway? Feb 15 14:43:24 <_law_> psokolovsky: only for testing.. so i have got ssh for udev work... Feb 15 14:43:29 pH5, yep, probably ;-) Feb 15 14:43:47 _law_, do that by unpacking your cpio.gz Feb 15 14:56:52 <_law_> psokolovsky: how can i cpio the whole files together? Feb 15 14:57:39 <_law_> psokolovsky: find * | cpio -o foo.cpio ? Feb 15 14:57:59 _law_, ah, worth man'ing as cpio is indeed not the most popular acrhiver Feb 15 14:58:23 _law_, yep, i beleive sth like that. Kernel doc on initramfs should have the sample command Feb 15 16:20:27 hrw|gone: In frustration, I took the DCF-660w card out and threw it on the ground Feb 15 16:20:30 now it works Feb 15 16:20:38 * jnc looks confused and amazed Feb 15 16:29:34 koen: Have you tried to build openmoko-dates today? Mine is still failing ( i have updated OE). But it seems like it is something else now. I'll do a pastebin Feb 15 16:30:45 ugh Feb 15 16:30:50 orinoco is messing this up Feb 15 16:30:55 I think Feb 15 16:32:42 http://pastebin.ca/357357 Feb 15 16:32:44 goxboxlive: https://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=189 and https://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=190 Feb 15 16:33:08 koen: I thought this bug was fixed a long time ago :( Feb 15 16:33:15 what bug? Feb 15 16:33:27 must have crept back in. orinoco_cs is picking up my wifi card Feb 15 16:33:39 jnc: that's hrw|gone's territory Feb 15 16:33:42 it should not. the correct behavior is for hostap_cs to pick it up Feb 15 16:33:43 ah okay Feb 15 16:33:49 will ping him when he returns, thanks Feb 15 16:34:07 goxboxlive: btw, your OE snapshot is too old :) Feb 15 16:34:24 goxboxlive: I already fixed the bug you are running into :) Feb 15 16:37:43 Is it realy to ols. "2007-01-13 15:27 org.openembedded.dev" So i have to start all over again, or can i keep my work directory? Feb 15 16:37:53 s/ols/old/ Feb 15 16:39:04 no, just 'mtn pull ; mtn up' Feb 15 16:40:22 I did that "2007-02-15 17:38 OE.mtn" There hasent been any changes since then. Feb 15 16:41:52 it should have a patch that patches out that header Feb 15 16:41:57 * koen <- food Feb 15 16:54:42 i get "compile-fix.patch is outdated" Feb 15 17:22:56 re Feb 15 17:23:15 jnc: pcmcia subsystem suxx when it comes to drivers matching Feb 15 17:23:37 jnc: card manufacturers suxx even more for not changing vendorid/productid Feb 15 17:25:12 hm Feb 15 17:25:23 hrw: unfixable? Feb 15 17:27:36 ipkg remove -recursive *orinoco* is simplest Feb 15 17:27:54 I need some time to check other way Feb 15 17:28:13 okay. yes I move the kernel drivers for orinoco out of the way, it works perfect now Feb 15 17:34:28 konqueror-embedded depends on libpcre, but it's not in its dependencies ... Feb 15 17:36:27 then add them Feb 15 17:53:03 _law_, so did it work? Feb 15 18:22:19 <_law_> psokolovsky: i?ll test it now.. Feb 15 18:47:22 <_law_> psokolovsky: ok i think i destroyed my image with cpio :-) where can i change this in OE? Feb 15 18:47:38 <_law_> where are the default devices build for /dev? Feb 15 18:49:48 _law_, grep -r ;-) Feb 15 19:07:58 hrw: Are you there/here? Feb 15 19:08:35 he is not here. Feb 15 19:08:45 Anyone hwo has managed to build openmoko-image yet? Feb 15 19:08:55 with OE Feb 15 19:09:55 Mine is failing with openmoko-dates, and i have done pull/update also deleted the sources and all *moko* in my work directory, but still it failes. Feb 15 20:06:55 :/ having trouble here setting /var and /usr to be located in a loopback ext3 filesystem, stored on the /media/hdd3 mount of a spitz Feb 15 20:09:46 how would I make the system mount the /media/hdd first, and then loopback systems after that? Feb 15 20:13:11 the entry in /etc/fstab isn't being read in order, or something Feb 15 20:13:20 'cause /dev/hda3 is being mounted in /media Feb 15 20:13:30 I thought it would be mounted in /media/hdd Feb 15 20:15:34 hmm. Feb 15 20:16:46 koen: what would you suggest to be the proper way in mounting some parts of the filesystem (/var, /usr) as loopback files on the vfat partition of hard disk with spitz... I am trying here, and it doesn't seem happy about it. Feb 15 20:17:07 the loopback stuff works, but it doesn't mount before the system boots up Feb 15 20:21:54 err Feb 15 20:22:11 correction, it looks like the MMC\SD card is being mounted on /media Feb 15 20:22:19 I must be misunderstanding something here Feb 15 20:22:44 when does /media/hdd get mounted? Feb 15 22:20:34 hi Feb 15 23:41:34 hey Feb 16 01:57:30 hello Feb 16 01:59:10 I' trying to compile angstrom, and it gets to qemu and complains Feb 16 01:59:11 | ERROR: "gcc" looks like gcc 4.x Feb 16 01:59:11 | QEMU is known to have problems when compiled with gcc 4.x Feb 16 01:59:11 | It is recommended that you use gcc 3.x to build QEMU Feb 16 01:59:11 | To use this compiler anyway, configure with --disable-gcc-check Feb 16 01:59:11 | FATAL: oe_runconf failed Feb 16 01:59:37 phire: Install gcc 3.3 on your host machine Feb 16 02:00:52 will it automatically recognize it? Feb 16 02:01:48 It should, yes, provided it's available as gcc-3.3, from memory Feb 16 02:02:35 cool Feb 16 02:06:17 NAbyss_: angstrom is unhappy that I put /usr and /var on other partitions, what do you think? Feb 16 02:07:09 jnc: Hmm, I can't see why, unless it's running out of space on one or the other; I don't know of any hard links that're created between the two Feb 16 02:07:36 NAbyss_: particularly there's rc scripts that cannot find executables from /usr when the thing boots Feb 16 02:07:56 I presume because it's not mounted Feb 16 02:07:56 jnc: Hmm.. is /usr mounted at the time? Feb 16 02:07:59 so my keymap gets all foobied up Feb 16 02:08:01 Yeah, that may be it Feb 16 02:08:11 You might have to shuffle the boot rc order in that case Feb 16 02:08:18 hmm Feb 16 02:08:34 I wanted to prevent root from filling up Feb 16 02:09:03 I may just have to make it Really Big (TM) Feb 16 02:09:03 jnc: I'd keep /usr on / for now, and just separate out /var Feb 16 02:09:06 ah Feb 16 02:09:11 will try that then, thanks Feb 16 02:09:30 /usr shouldn't change from day-to-day ops, only ipkg activity and other installation Feb 16 02:09:30 what's the biggest you think var would get, doing crazy stuff with ipkg Feb 16 02:09:38 1 gig? Feb 16 02:09:44 Hmm.. 1gb at the outside Feb 16 02:09:50 What device.. spitz? Feb 16 02:10:00 yep. borzoi specifically Feb 16 02:10:23 Yeah, unless you're running something insane, like hosting apache with a busy site on your Z.. 1gb should be more than enough Feb 16 02:10:56 I tried installing with a more-or-less stock partitioning scheme, and ran out of space trying to do an ipkg upgrade Feb 16 02:11:05 some file named 'perl' was like 98M Feb 16 02:11:22 ( in /var/lib/ipkg/something something) Feb 16 02:11:34 Yeah, I had to enlarge / on my Z; I think I went with 200mb root, 1gb /home and the rest in /media/hdd Feb 16 02:11:37 Whoa Feb 16 02:11:40 That sounds like a bug Feb 16 02:11:50 I can try again with a bigger partition Feb 16 02:12:26 with most common applications installed, what size would an all-in-one root partition be? Feb 16 02:12:52 I don't know, I keep thinking about debian how they say everything under the sun installed is 8-12G Feb 16 02:12:57 but this is pretty specific Feb 16 02:12:59 Let me grab my Z.. I've just got a basic gpe/abiword/gnumeric/gpsdrive setup on mine Feb 16 02:13:12 I'm going to have firefox, gimp, and a few dozen other apps like you mention there Feb 16 02:13:48 with firefox, gpsdrive, abiword, gnumeric and angstrom-gpe-image.. I have 125mb used Feb 16 02:13:57 So, 300mb should be a safe size Feb 16 02:13:57 it's been months since I used my Z. this is mostly because of bugs, but now I see a lot of the hardcore bugs are fixed. time to use it again :) Feb 16 02:14:08 oh wow. that's tiny. great Feb 16 02:14:16 Yeah, it's brilliant Feb 16 02:14:37 hey, do you know any devs who would want to borrow my remote to get it working for 2.6? Feb 16 02:14:53 I can loan it out for like 4-6 months or something Feb 16 02:15:26 Hmm, I'm not sure who'd be the person to contact about that.. post on the oe-dev mailing list? Feb 16 02:15:37 guess so Feb 16 02:15:52 the install process is different than I thought Feb 16 02:16:20 Personally, I'd see the place for it being an HID input/event driver in the kernel.. so oe-dev would probably hit more kernel devs than an angstrom user list Feb 16 02:16:36 ah Feb 16 02:17:03 But, IANAKernelDev or OEDev, so others would be more knowledgable than I :) Feb 16 02:18:26 do you do something like mount /dev/hda1 /hdd1; mkdir /hdd1/home; mount /dev/hda2 /hdd1/home; /mnt/card/gnu-tar zxvf hdimage1.tar.gz -C /hdd1 Feb 16 02:18:48 Nah, I just flash through the standard method Feb 16 02:18:49 I was complaining before about '/home/root' not being created on first boot, but it is there in the tarball Feb 16 02:18:53 oh Feb 16 02:18:59 updater.sh, hdimage1.tgz etc. on the CF card Feb 16 02:19:04 does the standard method format everything then Feb 16 02:19:18 Only the partitions you include the hdimageX file for Feb 16 02:19:28 i.e. hdimage1.tgz -> hda1, hdimage3.tgz -> hda3 Feb 16 02:19:30 oh! did not know that Feb 16 02:19:35 where is it documented? Feb 16 02:19:45 It isn't afaik, this's just what I've found from using it Feb 16 02:20:04 score. I learn something new from you everytime. Feb 16 02:21:35 NAbyss_: is updating process achieved by that japanese OK menu option 3? Feb 16 02:21:42 the pdf manual confuses me Feb 16 02:21:50 jnc: IIRC, it's hold down OK, 4, 1, Yes Feb 16 02:22:05 oh same as updating the kernel then Feb 16 02:22:56 Yeah Feb 16 02:24:33 qemu takes a long time to compile Feb 16 02:24:43 Yep, it's slow Feb 16 02:24:52 Once it's built though, it seldom changes Feb 16 02:25:04 finally just finished Feb 16 02:25:55 I put angstrom builds in a screen session Feb 16 02:26:00 and watch television. Feb 16 02:26:22 even with 4 xeon cpus at 3.06ghz, it takes about 9 hours to build Feb 16 02:26:34 Bloody hell Feb 16 02:26:54 I think it sometimes takes 5 hours Feb 16 02:27:01 but not remembering. Feb 16 02:27:26 NAbyss_: only 2 cpus actually, hyperthreading exposes 4 to linux Feb 16 02:28:31 It would be nicer if it didn't keep downloading files off ftp Feb 16 02:29:09 phire: Not many other places it can grab sources from Feb 16 02:29:17 for some reason, (I'm pointing my fingers at my isp) ftp transfers go at dialup speeds Feb 16 02:30:11 phire: That's.. weird Feb 16 02:30:19 phire: Is it only a certain source that does that? Feb 16 02:30:32 phire: You may be able to override the mirror used for given files Feb 16 02:30:40 NAbyss_: with the updater.sh 'hdimageX.tgz' trick, does it format hda3 as vfat -F32 if you have hdimage3.tgz ? Feb 16 02:31:00 no, everything, except http connections are limited to 6KB per connection Feb 16 02:31:21 phire: no QoS at the router? Feb 16 02:31:26 jnc: I'm not sure tbh Feb 16 02:31:38 phire: Wow, that's sucky.. can you complain to your ISP? Feb 16 02:31:58 he might be able to get around that by lowering the MTU size Feb 16 02:31:59 I could, but so is the entire country Feb 16 02:32:39 lowering MTU, what would that achieve? Feb 16 02:33:02 phire: Hmm, which package is slow ATM? Feb 16 02:33:14 phire: You should be able to select HTTP mirrors for GNU files and sourceforge Feb 16 02:33:42 glibc Feb 16 02:33:55 glibc-2.5.tar.bz2 Feb 16 02:34:20 phire: Let me grab the appropriate setting.. Feb 16 02:35:14 thats better, could I replace wget in the scripts with axel Feb 16 02:36:03 phire: In your local.conf, try this: export GNU_MIRROR = "http://www.planetmirror.com/pub/gnu" Feb 16 02:36:12 phire: That'll fix up GNU sources at least Feb 16 02:37:04 export DEBIAN_MIRROR = "http://mirror.optusnet.com.au/debian/pool" Feb 16 02:37:05 export SOURCEFORGE_MIRROR = "http://optusnet.dl.sourceforge.net/sourceforge" Feb 16 02:37:11 Those are two others that should work Feb 16 02:37:20 It's .au-centric, but there's plenty of bandwidth over the tasman Feb 16 02:38:25 yes Feb 16 02:48:33 phire: Did that have any success? Feb 16 02:49:25 kind of Feb 16 02:49:57 it still goes to the ftp, but if I press ctrl+c once it will try a mirror Feb 16 02:52:22 is there anyway I can replace wget with axel, which can go faster? Feb 16 02:53:24 phire: I'm not sure on that one; a hack would be to symlink wget to axel, alternately, you can play around with the .bb file that deals with wget Feb 16 02:53:33 k **** ENDING LOGGING AT Fri Feb 16 02:59:57 2007