**** BEGIN LOGGING AT Tue Aug 24 02:59:58 2010 Aug 24 08:14:24 lool, merci :) Aug 24 12:01:45 ogra: Why is my file system becoming fried when I install my own kernel on OMAP3 daily build? Aug 24 12:02:02 no idea, must be a kernel bug :P Aug 24 12:02:06 fsck from util-linux-ng 2.17.2 Aug 24 12:02:06 /dev/mmcblk0p2: The filesystem size (according to the superblock) is 533515 blocks Aug 24 12:02:06 The physical size of the device is 532153 blocks Aug 24 12:02:06 Either the superblock or the partition table is likely to be corrupt! Aug 24 12:02:11 which filesystem exactly ? Aug 24 12:02:11 Don't be silly Aug 24 12:02:19 ah, not the vfat Aug 24 12:02:24 Nope Aug 24 12:02:39 is that after or before installation ? Aug 24 12:03:13 i.e. after or before the autoreboot Aug 24 12:03:13 I dd, mount, put on my initrd-uimage-boot.scr, umount, boot Aug 24 12:03:42 I'll get you the full log Aug 24 12:03:42 you shouldnt have an fsck at all at that stage Aug 24 12:04:04 i dont get why you (or lool) even get one Aug 24 12:04:24 * lag *shrugs* Aug 24 12:04:28 the fsck should only happen after the partition table was newly written and the automatic reboot happened Aug 24 12:04:42 It's your toy, you tell me Aug 24 12:04:51 until the reboot there is no fsck at all Aug 24 12:04:59 Clearly there is Aug 24 12:05:10 I didn't know it was happening to lool too Aug 24 12:05:16 did it even reboot ? Aug 24 12:05:30 AFAIK it happens on first boot Aug 24 12:05:56 right, i hear you, *did it reboot at some point* Aug 24 12:05:56 first boot? you have 'fixrtc' set right in your bootargs? Aug 24 12:06:18 rcn-ee, isnt needed on first boot anymore Aug 24 12:06:26 only on subsequent ones Aug 24 12:07:10 http://paste.ubuntu.com/482869/ Aug 24 12:07:16 That's my boot.scr Aug 24 12:07:24 if it gets to an fsck before rebooting, then something with the reboot is wrong Aug 24 12:07:37 which in tunr points to either a busybox or kernel bug Aug 24 12:07:53 anyreason for no 'ro' in that? or isn't that needed anymore either? Aug 24 12:08:11 rcn-ee, it gets changed after first reboot Aug 24 12:08:24 jasper rewrites the kernel cmdline Aug 24 12:09:30 So there's nothing wrong with my boot.scr? Aug 24 12:10:07 jasper does: kick in in initramfs *before* anything gets mounted, then repartitions according to teh SD size, reformats the vfat completely, grows the rootfs to full size, touches the last mouted stamp on the FS, mounts it, does some setup and then reboots into oem-config Aug 24 12:10:33 *before* oem-config runs, the system does its first fsck Aug 24 12:10:41 http://paste.ubuntu.com/482871/ Aug 24 12:10:47 ah, cool.. Aug 24 12:11:41 ogra: Where does Jasper get it's kernel from? Aug 24 12:11:51 from the vfat Aug 24 12:12:05 "reformats the vfat completely" Aug 24 12:12:17 right for that part it pulls it from /boot Aug 24 12:12:39 So if I want to test another kernel, I have to put it in there too? Aug 24 12:12:44 the booting kernel is the uImage, the kernel after first reboot is a new uImage from /boot Aug 24 12:12:48 yeah Aug 24 12:12:57 Great! Aug 24 12:12:58 but your initrd seems strange in that log Aug 24 12:13:08 It's one I built Aug 24 12:13:24 aha Aug 24 12:13:29 What's wrong with it? Aug 24 12:13:32 its missing jasper completely Aug 24 12:13:44 Ahhhhhhhhhh Aug 24 12:13:49 you should see a lot of jasper messages after line 260 Aug 24 12:13:56 ogra: is omap4 in good state? Aug 24 12:13:59 You certainly know how to make my job difficult don't you? Aug 24 12:14:06 the first part of jasper is a local-premount script and is still very niosy Aug 24 12:14:11 asac, no Aug 24 12:14:19 asac, nothing is in a good state atm Aug 24 12:14:22 would anyone of you be able to test an image if we produced one at some point for us? Aug 24 12:14:31 just headless -> on sd ... booting -> done ;) Aug 24 12:14:36 not now ... in a few days Aug 24 12:14:46 telepathy-glib prevents image buiolding currently and oem-config is broken (upload pending) Aug 24 12:14:49 we dont have omap4 atm :(( Aug 24 12:14:50 asac: Yes Aug 24 12:15:05 asac, sure, no prob, just ping me with an url once you have iot Aug 24 12:15:16 good stuff ... Aug 24 12:15:24 ogra: could you give me the mkimage paramaters ;)? Aug 24 12:15:34 parameters ... damn my fingers never get it right Aug 24 12:16:48 ogra: What am I going to do about this? Is it just a matter of installing Jasper? Or are there other hidden nasties that you've implemented? Aug 24 12:17:09 asac, they are the same as for omap3 Aug 24 12:17:10 ogra: and if possible a good boot cmdline ;) Aug 24 12:17:32 ok good for the uimage Aug 24 12:17:34 lag, have a chroot with jasper in it and run update-initramfs there Aug 24 12:17:54 mkimage -A arm -O linux -T kernel -C none -a 0x80008000 -e 0x80008000 -n "Ubuntu Kernel" -d "$uboot_input_kernel" "$uboot_kernel" Aug 24 12:17:54 mkimage -A arm -O linux -T ramdisk -C gzip -a 0x0 -e 0x0 -n "Ubuntu Initrd" -d "$uboot_input_initrd" "$uboot_initrd" Aug 24 12:18:01 from the debian-cd script Aug 24 12:18:07 asac, ^^^ Aug 24 12:18:32 asac, and see lag's paste above for a boot.scr Aug 24 12:19:00 oh, wait, thats omap3 Aug 24 12:19:25 ogra: ok found it Aug 24 12:19:25 oh Aug 24 12:19:27 ;) Aug 24 12:19:36 looked really too similar ;) Aug 24 12:20:41 asac, http://paste.ubuntu.com/482874/ Aug 24 12:20:43 thats better Aug 24 12:21:06 ogra: How do you install Jasper? Aug 24 12:21:08 uboot_kernel_addr="0x80000000" Aug 24 12:21:08 uboot_ramdisk_addr="0x81600000" Aug 24 12:21:20 lag, apt-get install jasper in the chroot you use Aug 24 12:21:28 board_opts="mem=463M" Aug 24 12:21:29 ? Aug 24 12:21:36 yeah, needed Aug 24 12:21:38 thanks Aug 24 12:21:40 Tried that Aug 24 12:21:43 E: Unable to locate package jasper Aug 24 12:22:09 lag, its in main Aug 24 12:22:16 lag, only in maverick though Aug 24 12:26:21 * lag had an empty sources.list file Aug 24 12:27:02 Err http://uk.archive.ubuntu.com maverick/main armel Packages Aug 24 12:27:02 404 Not Found [IP: 194.169.254.10 80] Aug 24 12:27:05 Doh Aug 24 12:27:14 ogra: I told you that size of the fs issue would bite you in real life!! Aug 24 12:27:35 lool, no, it wouldnt if jasper would be executed Aug 24 12:27:55 lool, you didnt by chance keep a log from your broken boot test, did you ? Aug 24 12:29:27 ogra: Why isn't it finding the armel packages? Aug 24 12:29:41 http://194.169.254.10/ubuntu/dists/maverick/main Aug 24 12:29:43 ? Aug 24 12:30:38 lool, the point is that the initial rootfs is never used or touched, the whole thing gets adjusted before any fsclk, mount or other filesystem operations happen, while the corruption is ugly it *must* be gone after jasper has run Aug 24 12:30:56 so it shouldnt affect anyone unless jasper doesnt run Aug 24 12:31:18 lag, you want pool Aug 24 12:32:09 ? Aug 24 12:32:27 Instead of dists? Aug 24 12:32:28 ogra: I suspect resize2fs is running fsck or some equivalent Aug 24 12:32:39 ogra: I don't think it's the boot triggering fsck Aug 24 12:33:08 ogra: I'm not sure it's clear, but the issue seeems to indicate a corrupted to start with!! Aug 24 12:34:34 ogra: Aug 24 12:34:35 deb http://uk.archive.ubuntu.com/ubuntu/ maverick main restricted universe multiverse Aug 24 12:34:35 deb-src http://uk.archive.ubuntu.com/ubuntu/ maverick main restricted universe multiverse Aug 24 12:34:40 lool, jasper runs an fsck thats fully redirected to the logfile *right after* the new partitions were written Aug 24 12:35:19 lool, there is no way that you see *any* error message from that if jasper ran successfully ... and *after* jasper ran there shouldnt be any corruption anymore Aug 24 12:35:42 my point is that the corruption will not affect the actually installed filesystem Aug 24 12:36:19 i ageree that it needs to be fixed, but of the setup worked right you will never ever be able to even get any info about that corruption apart from jasper.log Aug 24 12:36:26 ogra: Are those sources.list entries correct? Aug 24 12:36:47 if you *see* an fsck during boot, something went completely wrong (i.e. jasper wasnt executed, you didnt use a 4G SD card etc) Aug 24 12:37:07 lag, no, needs to be ports.ubuntu.com Aug 24 12:37:14 k thanks Aug 24 12:37:18 the ports archive isnt mirrored Aug 24 12:37:27 lag, s|uk.archive.ubuntu.com/ubuntu|ports.ubuntu.com/ubuntu-ports| Aug 24 12:37:56 oh, right Aug 24 12:38:07 i missed /ubuntu-ports Aug 24 12:38:14 Rather, nobody has yet announced a public URL with a mirror. There exist some number of mirrors that aren't announced (I know of at least 4, but none better for the UK than ports.ubuntu.com) Aug 24 12:39:02 * ogra knows one Aug 24 12:39:16 but i guess there is a reason thats not announced anywhere Aug 24 12:39:30 ogra: It's in your house? Aug 24 12:39:41 I can't speak for anyone else, but I haven't announced mine because I don't serve it from a stable IP. Aug 24 12:40:07 lag, nah :) Aug 24 12:40:40 ogra: Aug 24 12:40:42 The following NEW packages will be installed: Aug 24 12:40:43 apt-xapian-index aptitude bc bogl-bterm consolekit cpp cpp-4.4 cryptsetup dbus defoma devio dmraid dosfstools ecryptfs-utils flash-kernel Aug 24 12:40:43 fontconfig fontconfig-config gcc-4.4-base gconf2-common gettext-base gir1.0-glib-2.0 hicolor-icon-theme hwdata iso-codes jasper keyutils Aug 24 12:40:43 language-selector-common laptop-detect libasound2 libatk1.0-0 libatk1.0-data libavahi-client3 libavahi-common-data libavahi-common3 Aug 24 12:40:44 libboost-iostreams1.42.0 libbsd0 libcairo2 libcanberra-gtk-module libcanberra-gtk0 libcanberra0 libcheese-gtk18 libck-connector0 Aug 24 12:40:47 libclass-accessor-perl libcroco3 libcups2 libcwidget3 libdatrie1 libdbus-glib-1-2 libdebconfclient0 libdebian-installer4 libdmraid1.0.0.rc16 Aug 24 12:40:52 libecryptfs0 libeggdbus-1-0 libept1 libexpat1 libffi5 libfontconfig1 libfontenc1 libfreetype6 libgconf2-4 libgcrypt11 libgdbm3 libgdk-pixbuf2.0-0 Aug 24 12:40:55 libgirepository1.0-1 libgmp3c2 libgnome-desktop-2-17 libgnutls26 libgpg-error0 libgssapi-krb5-2 libgstreamer-plugins-base0.10-0 libgstreamer0.10-0 Aug 24 12:40:58 libgtk2.0-0 libgtk2.0-bin libgtk2.0-common libgudev-1.0-0 libice6 libicu42 libidl0 libio-string-perl libiw30 libjasper1 libjpeg62 libk5crypto3 Aug 24 12:40:59 god ! Aug 24 12:41:01 libkeyutils1 libkrb5-3 libkrb5support0 libldap-2.4-2 libltdl7 libmpfr4 libnspr4-0d libnss3-1d libogg0 liborbit2 libpam-ck-connector libpango1.0-0 Aug 24 12:41:02 !paste Aug 24 12:41:03 use a pastebin ! Aug 24 12:41:04 libpango1.0-common libparse-debianchangelog-perl libparted0debian1 libpixman-1-0 libpolkit-gobject-1-0 libpython2.6 librsvg2-2 libsasl2-2 Aug 24 12:41:05 For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. Aug 24 12:41:07 libsasl2-modules libsigc++-2.0-0c2a libsm6 libstartup-notification0 libsub-name-perl libtasn1-3 libtdb1 libthai-data libthai0 libtiff4 Aug 24 12:41:10 libtimedate-perl libvorbis0a libvorbisfile3 libx11-6 libx11-data libxapian15 libxau6 libxcb-atom1 libxcb-aux0 libxcb-event1 libxcb-render0 Aug 24 12:41:13 libxcb-shm0 libxcb1 libxcomposite1 libxcursor1 libxdamage1 libxdmcp6 libxext6 libxfixes3 libxfont1 libxft2 libxi6 libxinerama1 libxml2 libxrandr2 Aug 24 12:41:16 libxrender1 lsof make oem-config oem-config-debconf os-prober parted perl perl-modules python-apt python-cairo python-dbus python-debian Aug 24 12:41:19 python-gobject pyt Aug 24 12:41:23 Argggggggggggggggggggggggggggggggggggggggggggggggggggg Aug 24 12:41:31 Paste bin is not as much fun! Aug 24 12:41:40 heh Aug 24 12:41:49 Yeah, but not using one reliably makes me go fiddle with IRC commands :p Aug 24 12:41:52 anyway, jasper depends on oem-config indeed Aug 24 12:42:05 persia: I noticed Aug 24 12:42:08 ogra: Again, I'm not sure you see that the problem is *earlier* Aug 24 12:42:09 the other deps you see are all oem-config deps Aug 24 12:42:21 ogra: I'm pretty sure the issue is that the fs in the image itself is truncated Aug 24 12:42:36 whatever you do starting from there is bound to suffer from the corruption Aug 24 12:42:38 lool, well, but where did your fsck come from ? Aug 24 12:43:04 there is no technical possibility that you ever see any output of fsck if jasper ran Aug 24 12:43:36 ogra: My understanding is that jsaper does a resize2fs Aug 24 12:44:20 lool, jasper runs completely silently *all* output of any subcommands is redirected to a logfile Aug 24 12:44:31 Well I still got some output apparently Aug 24 12:44:45 if jasper ran there cant be any discrepancy between the partitioning and the fs anymore Aug 24 12:44:47 Note that your pipe + read might not be enough to stop commands from writing to /dev/tty Aug 24 12:45:07 What type of bot is ubot2? Aug 24 12:45:09 since the rootfs partition was resized (no matter of the corruption or not, sfdisk doesnt care) Aug 24 12:45:55 lool, well, then someone must have secretly uploaded a changed jasper :) we had various bugs and breakages that caused fsck errors that only show up in the log Aug 24 12:46:57 i wouldnt know why that suddenly would change and ignore a redirect of stderr and stdout Aug 24 12:47:40 Well it's just a guess Aug 24 12:47:54 well, my guess is that jasper didnt run at all for you Aug 24 12:47:56 either the fs is corrupted in the image, or it gets corrupted in the resize Aug 24 12:49:05 ogra: http://people.canonical.com/~lool/fsck.png Aug 24 12:49:35 lool, there is no initramfs output at all Aug 24 12:49:42 ogra: It's earlier Aug 24 12:49:49 ogra: But I can't scroll back Aug 24 12:49:52 do you have a shot of that too ? Aug 24 12:49:54 Let me try with serial console Aug 24 12:49:55 ah, darn Aug 24 12:50:12 ogra: is there a way to turn serial console output during boot? Aug 24 12:50:25 only by changing boot.scr on the vfat Aug 24 12:51:03 as long as plymouth breaks once console= is set we cant make it a default else we wont have any splash anymore Aug 24 12:51:18 (second boot is completely with splash etc) Aug 24 12:52:56 ogra: Can you give me the commands to run to type in u-boot? Aug 24 12:53:04 one sec Aug 24 12:53:06 I don't have a boot.scr nearby Aug 24 12:53:27 fatload mmc 0:1 0x80000000 uImage Aug 24 12:53:32 fatload mmc 0:1 0x81600000 uInitrd Aug 24 12:53:32 So I'm on the serial line Aug 24 12:53:39 ** Unable to use mmc 0:1 for fatload ** Aug 24 12:53:43 mmc init Aug 24 12:53:47 sh, indeed Aug 24 12:53:52 s/sh/ah/ Aug 24 12:53:57 Ok, this is done Aug 24 12:54:00 Now the tricky part Aug 24 12:54:06 setenv bootargs vram=12M omapfb.mode=dvi:1280x720MR-16@60 console=ttyS2,115200n root=/dev/mmcblk0p2 fixrtc Aug 24 12:54:11 bootm 0x80000000 0x81600000 Aug 24 12:54:23 thanks, booting Aug 24 12:54:35 ogra: !paste Aug 24 12:54:38 Lol Aug 24 12:54:42 :D Aug 24 12:54:44 :P Aug 24 12:54:45 I get kernel output on serial console at least Aug 24 12:55:45 ogra: So now I have Jasper installed, when I run my usual script to build an initrd, Jasper will just appear in there yes? Aug 24 12:56:17 if jasper is inside the chroot you run upadte-initramfs in it will end up inside the initrd, yes Aug 24 12:56:37 Jasper is installed in the chroot, yes Aug 24 12:56:54 well, then it should end up in your initrd Aug 24 12:57:05 * lag crosses fingers Aug 24 12:58:08 ogra: http://paste.ubuntu.com/482895/ Aug 24 12:58:38 lool, it doesnt reboot after line 30 ? Aug 24 12:58:55 ogra: no Aug 24 12:58:57 aha Aug 24 12:59:24 thats one of the issues, can you try booting a second time ? Aug 24 12:59:32 that will behave differently Aug 24 12:59:37 right Aug 24 13:00:19 ogra: I'm quite busy for doing long testing; would you be interested in looking into it with qemu? Aug 24 13:00:24 it's not hard to reproduce in qemu Aug 24 13:00:24 lool, also /var/log/jasper.log would be intresting Aug 24 13:00:52 well, using these images in qemu will require a lot of tinkering with images in advance Aug 24 13:01:03 ogra: Why is that? Aug 24 13:01:24 because the FS needs to actually be resized Aug 24 13:01:38 if you just boot the img file there is no spare space Aug 24 13:01:50 Yes, which exposes this bug which apparently other people are seeing Aug 24 13:02:04 so you would need to create a 4G image and dd the img file into it to have the proper way Aug 24 13:02:17 As I noted in email, if I append some zeroes at the end of the image, I don't get this issue Aug 24 13:02:22 like 10 MBs or so Aug 24 13:02:22 yes, the bug is on antimony side though Aug 24 13:02:45 ogra: It's trivial to add space to the QEMU SD backing image... Aug 24 13:02:51 dd if=/dev/zero bs=1024 count=10240 >> /tmp/maverick-preinstalled-netbook-armel+omap.img Aug 24 13:03:03 but it shouldnt appear if there is space to resize to, the resizing "fixes" the filesystem Aug 24 13:03:33 thats why we dont see it (apart from jasper.log) on real installs Aug 24 13:03:45 Has the meeting been cancelled? Aug 24 13:03:49 lag, nope Aug 24 13:04:08 ogra: Ok; I have other things to work on and I'm not blocked by that bug; I found it relevant to share what I believe is a bug in the Ubuntu images exposed by testing in QEMU, but it's time consuming for me to make my point and I have tons of other things to look into, sorry Aug 24 13:04:31 lool, yeah, understood, thanks for pointing it out Aug 24 13:04:32 I can share the recipe to reproduce, but you don't seem interested in resolving what appears to me as a bug in the jasper scripts Aug 24 13:04:44 its not in jasper Aug 24 13:04:57 the FS is apparently corrupt without resizing Aug 24 13:05:08 which means its corrupt at creation time Aug 24 13:05:31 and thats surely something i'll look at Aug 24 13:05:45 what is worrying me more though is that the reboot doesnt work Aug 24 13:15:11 ogra: When I build my own initrd, do I have to execute depmod at any stage? If so, what stage? Aug 24 13:15:21 ogra: Well I'm not entirely sure it's corrupt at image creation time Aug 24 13:15:26 update-initramfs does that iirc Aug 24 13:15:36 ogra: jasper resizes the partitions before calling resize2fs, to it might get that part wrong... Aug 24 13:15:38 lool, only your jasper.log could tell :) Aug 24 13:15:58 but after you sent me that mail i checked a jasper.log and i see corruption there Aug 24 13:16:27 ogra: Well perhaps you can reproduce in QEMU and dig the info you need? It's not hard, the .debs are pre-built, wont mess your other qemu, and you just need a maverick daily image to try out Aug 24 13:17:06 lool, i will Aug 24 13:18:55 NCommander: I've just pasted the bug id on #linaro, but you don't seem to be in there Aug 24 13:19:00 NCommander: https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/623375 Aug 24 13:19:02 Launchpad bug 623375 in initramfs-tools (Ubuntu) "Skipping the bootloader installation when creating rootfs or installation media (affects: 1) (heat: 8)" [Undecided,New] Aug 24 13:19:33 lool, i pointed him to it in the arm meeting Aug 24 13:20:16 lool: yeah, I saw that. Whoops on my part, though ideally, we shouldn't directly run update-initramfs in a chroot either Aug 24 13:24:14 NCommander: I don't see an issue with running update-initramfs in a chroot, in fact it's the only way to generate an initrd Aug 24 13:24:38 ogra: ? Aug 24 13:24:56 lag, ? Aug 24 13:25:05 Look up Aug 24 13:25:08 update-initramfs does that iirc Aug 24 13:25:09 lool: I meant run it if a trigger calls it; we should run it if the user calls it directly. Obviously the generic bit of generic subarch detection is a bit too generic :-) Aug 24 13:25:14 look up too ;) Aug 24 13:25:23 NCommander: I don't think we want to run flash-kernel though Aug 24 13:25:27 Which is what the bug is about Aug 24 13:25:48 lool: I think we can detect the chroot via a syscall (I remember there's a way to do it in C), and then we can adjust what we do based off that Aug 24 13:25:52 lool, before NCommander changed it it wouldnt have executed :) Aug 24 13:25:55 lag: update-initramfs does that iirc Aug 24 13:25:57 :) Aug 24 13:26:25 lool, https://blueprints.edge.launchpad.net/ubuntu/+spec/mobile-maverick-arm-improved-subarch-detection Aug 24 13:26:43 ogra: right, since flash-kernel --supported would have ailed Aug 24 13:26:45 *failed Aug 24 13:26:49 * NCommander needs a new keyboard Aug 24 13:28:40 NCommander: it's a bad idea to detect the chroot Aug 24 13:28:58 NCommander: d-i and ubiquity will install into /target, where you do want flash-kernel to run Aug 24 13:29:31 lool: we can have special logic in flash-kernel-installer to simply manually call flash-kernel and work around the problem. Aug 24 13:30:01 but point taken Aug 24 13:30:12 lool: what do you suggest for a solution? Aug 24 13:31:23 ogra: http://paste.ubuntu.com/482907/ Aug 24 13:31:59 lag, cool ! Aug 24 13:32:11 How is that cool? Aug 24 13:32:21 it works Aug 24 13:32:31 ogra: FATAL: Could not load /lib/modules/2.6.35-18-omap/modules.dep: No such file or directory Aug 24 13:33:20 ignore that, you get to ubiquity-dm Aug 24 13:33:37 (which doesnt start because of a bug in oem-config) Aug 24 13:33:51 I don't want to ignore that Aug 24 13:33:58 I want to fix that :) Aug 24 13:36:40 ogra: -^ Aug 24 13:45:53 NCommander: well it was discussed in the bug already Aug 24 13:46:10 lool: I haven't had a chance to read it. LP from China is ... interesting Aug 24 13:46:39 NCommander: Oh you're there already, how is it? Aug 24 13:46:43 Chinese? Aug 24 13:47:10 lost of meo-burgers :) Aug 24 13:47:10 lool: I've been here for a little over a week and a half :-) Aug 24 13:47:12 *meow Aug 24 13:47:19 ogra: that's vietnam Aug 24 14:00:34 ian_brasil: so, I admit, I haven't looked at the liquid stuff in-depth. Do you have a seed yet? Aug 24 14:00:40 yes Aug 24 14:00:43 (or a metapackage, but a seed is preferable) Aug 24 14:00:45 oooh Aug 24 14:00:47 handy Aug 24 14:00:55 ogra: you really should watch food network sometime. You'd know that in china, dog is the meat of choice. Aug 24 14:01:16 ian_brasil: so if I build a chroot, and then do apt-get install kubuntu-liquid^, everything works? Aug 24 14:01:31 no Aug 24 14:01:47 ian_brasil: it doesn't? Aug 24 14:02:00 * NCommander admits he's not sure what your seed name is Aug 24 14:02:00 no ..we dropped the liquid name Aug 24 14:02:08 liquid was just a code name Aug 24 14:02:26 ian_brasil: oops, I'm out of the loop >.<; Aug 24 14:02:43 damn..fire alarm..back in a minute Aug 24 14:14:26 excitement over Aug 24 14:14:35 NCommander: the meta package is kubuntu-mobile Aug 24 14:14:49 https://edge.launchpad.net/ubuntu/maverick/+package/kubuntu-mobile Aug 24 14:15:25 ian_brasil: heh Aug 24 14:15:35 ian_brasil: right, next step is to modify livecd-rootfs to add it as an image target Aug 24 14:16:11 ian_brasil: its pretty straightforward on where the edits have to be made. Only thing is you want to pull by task, and not metapackage Aug 24 14:16:26 ok Aug 24 14:18:04 we have done that before i seem to remember Aug 24 14:20:41 NCommander: i will ping you when that is done Aug 24 14:21:21 ogra: Hi! I was wondering what OMAP3 boards you support in your OMAP prebuilt images. Just the Beagle board? Aug 24 14:21:38 cureently just the beagle, yes Aug 24 14:21:46 ian_brasil: thanks. I can't merge or do the upload, but I can guide you with the edits and doing some initial testing Aug 24 14:22:04 ogra: all right, thanks! Aug 24 14:22:11 mopdenacker, iegp2 might work if you have a special u-boot, gumstix too Aug 24 14:22:18 ootb only beagle though Aug 24 14:22:59 mopdenacker: ogra: xm should also work, but we have a mmc issue Aug 24 14:23:12 err, indeed, i counted that as beagle :) Aug 24 14:23:25 ogra: there are just too many different beagles at the marked hah Aug 24 14:23:34 b, c, XM Aug 24 14:23:35 C4 and XM are the ones we'll support by release day Aug 24 14:23:41 sure Aug 24 14:23:49 b will likely have issues to even boot Aug 24 14:23:57 hehe, probably Aug 24 14:23:58 at least our current images Aug 24 14:24:16 unless we get a minimal image, no way to support it Aug 24 14:24:25 even for simple boot Aug 24 14:24:28 right Aug 24 14:24:45 mine is running well with a simple rootstock fs Aug 24 14:24:45 ogra: Any chance of building an alt-inst image? Aug 24 14:24:54 GrueMaster, nope Aug 24 14:25:01 ok. Aug 24 14:25:18 ogra: I commented at the ureadahead bug, it doesn't make any difference for us Aug 24 14:25:22 I checked that Aug 24 14:25:24 GrueMaster, we are fully focused on preinstalled images and would have no way to support systems without NAND in a sane way atm Aug 24 14:25:30 but let's wait for the new rewrite Aug 24 14:25:50 and it doesn't make any sense to run until we get OOM, doesn't help at all Aug 24 14:25:51 rsalveti, yeah, i talked to Keybuk already this morning (thats why he commented on the bug) Aug 24 14:25:52 ogra: ok. Aug 24 14:26:06 ogra: I saw that, and at the time I got to irc he wasn't there anymore Aug 24 14:26:11 GrueMaster, would be good to probably address that in N Aug 24 14:26:27 oh, he dropped again ... ? bah Aug 24 14:26:30 ogra: What about a preinst-min image? Aug 24 14:26:43 ogra: so let's wait for the new version, then we can test Aug 24 14:26:47 GrueMaster, requires time i dont have Aug 24 14:26:47 Minor mods needed. Aug 24 14:26:53 but I doubt it will make any difference Aug 24 14:26:54 no, massive mods needed Aug 24 14:26:56 Ok, just a suggestion. Aug 24 14:27:01 thats the prob Aug 24 14:27:12 but will be happy to be wrong Aug 24 14:27:24 there apparently is a base image already that has been implemented for special purposes Aug 24 14:27:43 to change that requires a ton of changes Aug 24 14:28:13 (it doesnt build an image, just a base tarball for local testing purpose) Aug 24 14:29:01 GrueMaster, i'd love to have it, but the current quality of the images we actually have makes me thing that i wont find the time Aug 24 14:29:07 *think Aug 24 14:29:14 * ogra gest some coffee Aug 24 14:29:34 Actually, if we had them, it might be easier to debug netbook image issues. Aug 24 14:30:14 And daily image testing could continue at a limited pace. Aug 24 14:30:35 * GrueMaster remembers making this same arguement months ago. Aug 24 14:33:32 so I have a new imx51 "SmartTop" system sitting here I want to upgrade to maverick. Will it break ? Aug 24 14:35:49 GrueMaster, yes, and i said "i'd love to have them but i dont know if i have the time" Aug 24 14:36:10 rsavoye, no idea, we dropped imx51 support with lucid Aug 24 14:36:45 bummer. Can I upgrade and leave the kernel alone ? Aug 24 14:36:54 mater of luck i guess Aug 24 14:36:55 it's running karmic now Aug 24 14:37:45 rsavoye: You'd need to bump the kernel. Karmic was Armv6+vfp. Aug 24 14:38:09 Might be as easy as copying the config and rebuilding. Aug 24 14:38:16 it might be possible to run v7 userspace with a v6 compiled kernel, not sure Aug 24 14:38:35 as long as your userspace is all v7 Aug 24 14:38:54 i'D just keep a backup of the original system and try it :) Aug 24 14:39:02 so to do the user space hack, I'd just change sources.list? Aug 24 14:39:07 yes Aug 24 14:39:19 guess I'll let you know :-) Aug 24 14:39:23 but really, keep a backup ! Aug 24 14:39:35 it's fresh hardware, so nothing to loose Aug 24 14:40:03 GrueMaster, our buildds were all running v5 compiled kernels until we got the babbage boards in the DC Aug 24 14:40:25 there were issues but i dont know if they were caused by v5 vs 6/7 Aug 24 14:42:21 is ther an easy way to block upgrading the kernel when I upgrade everything else ? Aug 24 14:42:46 it shuldnt upgrade the kernel unless you already have an ubuntu kernel package Aug 24 14:42:51 which i doubt Aug 24 14:43:04 cool, I'll give it a try Aug 24 14:43:26 rsavoye: I confirm. I use Lucid on my IGEPv2 board, and everything gets updated, except my custom kernel. Aug 24 14:43:33 rsavoye: YOu can try to do a dist-upgrade, but cancel when it asks to continue. Then do a manual install on individual packages. Aug 24 14:45:21 dist-upgrade or update-manager -d ? Aug 24 14:45:35 both should work Aug 24 14:45:50 or do-release-upgrade Aug 24 14:46:03 (the cli version of update-manager) Aug 24 14:49:34 ah, I'll try that Aug 24 14:54:55 ogra: did you solve your issue with the latest preinstalled images? I'm asking because we need to make our own for our internal releases, and I need your howto on this... Aug 24 14:55:34 mopdenacker, half of it Aug 24 15:01:45 ogra: thanks for the status update. Good luck with the second half! Aug 24 15:02:32 mopdenacker, second half is a package failing to build which keeps us from rolling images atm we're trying to get that fixed asap Aug 24 15:06:45 hey devilhorns Aug 24 15:06:51 hi ogra :) Aug 24 15:06:52 just sent a mail your way :) Aug 24 15:06:56 oh super :) Aug 24 15:07:04 guess yesterday was just bad timing :) Aug 24 15:07:10 yeah Aug 24 15:07:16 hehehe Aug 24 15:07:19 which TZ are you in ? Aug 24 15:07:26 * ogra is in europe Aug 24 15:07:34 EST Aug 24 15:07:45 in america Aug 24 15:07:52 east coast Aug 24 15:08:02 hi, what kind of hardware can you use ubuntu on? Aug 24 15:08:17 can you give me a system example? Aug 24 15:08:40 ah, east coast is fine, the time you replied looked more like asian morning to me :) Aug 24 15:08:43 ogra, k, just got the mail Aug 24 15:08:47 great Aug 24 15:08:54 davidm, meet devilhorns Aug 24 15:08:57 :) Aug 24 15:09:03 bench_adidas: omap3, omap4, i.mx51 Aug 24 15:09:03 rsalveti: I sent you a uImage.. Aug 24 15:09:06 ogra, well, was up late last night talking to raster :) and working on some code Aug 24 15:09:17 it will either not boot at all, work perfectly, or something in between ;-) Aug 24 15:10:07 hi davidm :) Aug 24 15:10:37 can i get a system example please? Aug 24 15:11:28 bench_adidas, C4 beagle board will work with 10.10, marvell dove and freescale babbage too Aug 24 15:11:34 is there one? an example or is this topic the new sh!t Aug 24 15:11:38 err 10.04 Aug 24 15:11:49 ?... Aug 24 15:12:39 ah! cool! Aug 24 15:13:23 the dove isnt on sale and the babbage is really expensive Aug 24 15:13:49 ogra, typing up a response now wrt that email :) Aug 24 15:13:57 devilhorns, awesome :) Aug 24 15:15:15 babbage is named after a character eh! Aug 24 15:15:43 likely, ask freescale marketing dept. :) Aug 24 15:16:04 OMAP3 processor <<< only uses 2 watts? Aug 24 15:16:14 its ARM, sure :) Aug 24 15:17:17 ogra, r u supposed to be a computr architech? Aug 24 15:17:31 me ? Aug 24 15:18:16 oh-gra! Aug 24 15:18:31 i'm a software integrator ... Aug 24 15:18:46 ive never herd of software inegrator Aug 24 15:18:50 stealing work from others and making OSes out of them :) Aug 24 15:18:56 lol Aug 24 15:20:02 hello devilhorns Aug 24 15:20:09 hi davidm :) Aug 24 15:20:39 I'll look forward to your response to ogra's email Aug 24 15:20:49 :) about to send it now Aug 24 15:27:37 ogra, in my day, we called that Systems Integration :) Aug 24 15:28:09 yeah, probably the better term :) Aug 24 15:28:22 :) Aug 24 15:54:54 bye channel. thanks Aug 24 16:44:47 robclark: thanks, will try now Aug 24 16:44:51 was at lunch :-) Aug 24 16:44:57 cool Aug 24 16:50:26 GrueMaster: and your es2, how is it going? Aug 24 16:51:22 Had to clean up a large area of my office yesterday to accommodate it along with my XM & Dove. Just getting the final wiring in place now. Aug 24 16:51:49 I did manage to get it booting to X and netbook-launcher-efl Friday. Aug 24 16:52:33 cool Aug 24 16:52:43 hopefully we're getting ours next week Aug 24 17:00:03 hope your XM board works better than mine... Aug 24 17:01:40 mine only works with mem=256M :-) Aug 24 17:01:50 but it's a P8 Aug 24 17:12:25 oh well, looks like I just wedged my imx51 board :-( Time to start over Aug 24 17:13:02 rsavoye: if you got the kernel and the boot loader working, I'd suggest you to generate a rootfs with rootstock Aug 24 17:13:15 just a minimal, so you can get something to work with maverick Aug 24 17:13:27 it hangs after "init crypto disks" Aug 24 17:13:37 maverick? Aug 24 17:14:08 this was after upgrading to maverick user space on the karmic kernel Aug 24 17:14:11 rsavoye, on a serial console ? Aug 24 17:14:24 you might not have serial getty enabled Aug 24 17:14:30 (doesnt happen by default) Aug 24 17:14:32 hdmi monitor Aug 24 17:14:43 ah, k Aug 24 17:14:52 I guess I should pop the cover off the SmartTop an insatll the serial dongle Aug 24 17:15:28 well, it should work on a monitor for sure Aug 24 17:15:43 that's what I hoped Aug 24 17:15:59 also its weird that you have crypto disks installed Aug 24 17:16:00 I get a bunch of boot messages then it hangs Aug 24 17:16:15 I don't think I did, not sure where that came from Aug 24 17:16:22 i think thats only kept by the installer if you actually chose to encrypt your homedir Aug 24 17:16:33 al I did was an upgrade :-) Aug 24 17:16:39 well, its an upgrade Aug 24 17:16:43 no user dir beyond the default one Aug 24 17:16:49 *someone* picked to install it :) Aug 24 17:17:10 poltergists are hell... Aug 24 17:17:19 hehe Aug 24 17:18:18 robclark: well, the display works :-) Aug 24 17:18:37 robclark: can see 2 penguins and the boot log Aug 24 17:18:53 but got stuck after udev Aug 24 17:19:07 but I only got your uImage, so could be missing modules and stuff Aug 24 17:19:30 robclark: which tree is this one? Aug 24 17:31:59 * prpplague looks around for a blue box Aug 24 17:35:09 ol Aug 24 17:36:29 err lol Aug 24 17:36:35 dunno what happened to the first L there Aug 24 17:37:48 ogra, around ? Aug 24 17:41:59 ogra, ok, well when you get this, could you email me a list of packages that I will need to focus on ? Obviously the standard EFL libs and such, but not sure what other ones there may be. Thanks :) Aug 24 17:59:28 rsalveti: prpplague found something interesting.. Aug 24 17:59:42 I'd been enabling only one 1fb.. but when 2nd fb is enabled then it doesn't work Aug 24 17:59:51 so I think we have something to go on now Aug 24 17:59:59 hm, ok Aug 24 18:00:18 rsalveti: all of th e patches work, but when only 1 fb is enabled Aug 24 18:00:25 rsalveti: try with your existing kernel tree.. and change # of fb's to 1.. which I think you can even do w/ a bootarg Aug 24 18:00:27 rsalveti: you can change the .config to turn those off Aug 24 18:02:43 prpplague: robclark: interesting, will try Aug 24 18:20:15 prpplague, rsalveti: I see what the issue is.. will have a fix in a bit Aug 24 18:20:32 basically fb1 gets updated with new size, instead of fb0 Aug 24 18:21:00 but I'm not completely sure if it is the same issue that rsalveti sees.. with this issue, you'd still see scrambled picture, rather than no picture Aug 24 18:21:22 brb Aug 24 18:24:02 robclark: well it depends on the display, some displays won't even display the scrambled picture to do incorrect data Aug 24 18:28:17 prpplague: hmm.. from display standpoint, the signal timings should be good.. so from monitor's point of view the picture is just fine.. the user is just looking at some abstract art ;-) Aug 24 18:30:53 (well, ok, you could run into some issues switching from lower to higher resolution, if you've exceeded the size of your framebuffer) Aug 24 18:31:43 robclark: so you found the code issue? Aug 24 18:32:01 yeah.. basically the wrong framebuffer gets resized Aug 24 18:32:15 should be fixable Aug 24 18:32:37 robclark: what does it do, just assume it is the last framebuffer available? Aug 24 18:33:02 seems to be Aug 24 18:33:34 makes sense Aug 24 18:34:29 yeah, basically each fb shares same dssdev instance.. so when second fb registers it overwrites the first fb's callback Aug 24 18:36:15 cool, nice catch Aug 24 18:50:58 hi Aug 24 18:51:13 do you have accelerated X on omap3 already? Aug 24 18:51:52 Baybal: you have the sgx package, but it's not yet integrated at X11 as we have for n900 Aug 24 20:23:24 rsalveti: did you get a chance to test the kernel with 1fb? Aug 24 20:23:30 mopdenacker: greetings Aug 24 20:23:57 prpplague: not yet, needed to debug another issue Aug 24 20:24:01 will do in some minutes Aug 24 20:24:37 rsalveti: np, just checking Aug 24 20:26:53 prpplague: hi! Aug 24 20:57:22 ogra_cmpc: Still up? Aug 24 22:07:09 rsalveti: what do you mean by 'we have the sgx package on omap3'? do you mean a ubuntu package or the TI tar ball? Aug 24 22:07:40 ndec: just tar ball, we're still waiting for the official deb :-) Aug 24 22:07:57 rsalveti: too bad ;-) that would have been better for me ;-) Aug 24 22:08:17 haha :-) Aug 24 22:12:54 prpplague: robclark: sorry for taking so long, but I can confirm that it works when I compile the kernel with just one framebuffer Aug 24 22:13:06 with robclark's patches on top of ubuntu kernel Aug 24 22:13:20 rsalveti: dandy Aug 24 22:16:40 oh goody :-) Aug 24 22:16:48 * robclark is working on better patch to support multi fb Aug 24 23:55:25 Weird... trying to use a pl2303 usb-serial ON the beagleboard... and it's giving me garbage. Aug 24 23:55:27 pl2303_process_read_urb - tty_flag = 0 Aug 24 23:56:01 H�H�H **** ENDING LOGGING AT Wed Aug 25 02:59:57 2010