[05:45] - Linux >= 2.6.20 with I/O accounting support (CONFIG_TASKSTATS, CONFIG_TASK_DELAY_ACCT, CONFIG_TASK_IO_ACCOUNTING): Not found === hrw|gone is now known as hrw [08:13] morning [08:14] hi hrw! [08:21] morning .pl [08:26] amitk, hey, how are you [08:27] good, how is the flooding situation in .pl? [08:37] amitk, it roughly moved on to germany now [08:38] ogra: ohh? [08:39] the oder flows from poland through germany [08:39] i think the highest levels are done in poland and move down the river now [08:42] ogra: Poland is big country - flood to my city has not yet arrived [08:42] oh, wow [08:42] but Szczecin is prepared for Odra flood [08:42] we have lot of space for water where no one is allowed to live [08:43] german news said the hump would arrive at the german boarder last night [08:43] and bigger problem is not water from south but from returning water from Baltic Sea [08:43] Szczecin tomorrow/Saturday iirc [08:43] yeah [08:44] oh, right, Szczecin comes after it went through germany [08:47] zyga: sent [08:49] hrw, did you get a bounce? [08:49] hrw, I suspect you did what I did some time ago [08:50] zyga: I just sent it again though my own smtp [08:50] hrw, I think you have to send it thru google's own SMTP using "that" domain's account to actually reach your recipients [08:51] brb [08:51] zyga: smtp is smtp - as long as email is delivered [09:00] back [09:01] hrw, yeah but google seemed to block my other address, anyway the message is away [09:01] hrw, I wonder what people think [09:01] hrw, -> (other chan) [09:33] ndec, https://wiki.ubuntu.com/MaverickReleaseSchedule [10:01] hmm, http://paste.ubuntu.com/440283/ wonder what is happening at line 595 for nfsroot [10:02] ogra: ^ any ideas [10:05] amitk, i remember there were mountall issues with nfsroot, not sure they are already solved [10:05] i guess we need to talk to Keybuk [10:40] amitk, https://bugs.edge.launchpad.net/ubuntu/+source/mountall/+bug/537133 [10:40] Launchpad bug 537133 in portmap (Ubuntu Lucid) (and 3 other projects) "mountall issues with NFS root filesystem (affects: 10) (dups: 2) (heat: 72)" [Medium,Fix released] [10:41] ubot2, liar !! [10:41] Factoid 'liar !!' not found [10:41] (its not fix released for mountall, only for portmap) [10:42] ogra_cmpc: hmmm... not sure if the second issue is similar to min. [10:42] *mine [10:44] there is also https://bugs.edge.launchpad.net/ubuntu/+source/mountall/+bug/578851 [10:44] Launchpad bug 578851 in mountall (Ubuntu) "Fail to remount root on nfsroot install (affects: 1) (heat: 6)" [Undecided,New] [10:45] though you dont seem to have /dev/nfs in your fstab [10:50] amitk, erm, do you use an initramfs ? [10:50] doesnt look like it [10:50] ogra_cmpc: no [10:50] i dont think nfs support is compiled in [10:50] ogra_cmpc: it is [10:50] hmm [10:51] and /dev/nfs is in my /etc/fstab for the remote root fs [10:51] did you try with initramfs ? [10:51] i bet it would work [10:54] ogra_cmpc: no, and to be honest I don't want to. I want to be able to boot-test several different arm boards. Compile kernel and copy to webserver. And the board gets its rootfs over nfs [10:55] there is not .deb packaging around these kernels, straight mainline + patches [10:55] ootserver=192.168.0.254, rootserver=192.168.0.101, rootpath= [10:55] seems your rootpath doesnt get handed over properly [10:56] line 454 [10:57] try with explicitly setting rootpath=/shared/nfs [10:57] and see if IP-config gets it then [10:58] else i'd actually say its an IP-Config bug [10:58] indeed, good catch [10:59] also is your export in the server side proper ? (try to mount it rw from somewhere else) [11:00] ogra_cmpc: yeah tried that [11:00] while rootpath isnt set it obviously gets to mountall, so it must have acessed it [11:00] ogra_cmpc: it should get the rootpath or equivalent from the kernel cmdline [11:01] (though i dont know if mountalal doesnt try to grab rootpath ... if its empty it likely doesnt know what to do) [11:01] *mountall [11:02] well, in initramfs we have a special script that parses nfsroot and actually exports rootserver and rootpath, might be that mountall relies on that [11:03] see /usr/share/initramfs-tools/scripts/nfs [11:19] amitk, heh, do you have devtmpfs support compiled in ? [11:20] ogra_cmpc: yes, I saw that mount option for /dev and added it in last night :) [11:20] i dont see it listed between 461 and 484 [11:20] doesn't ltsp use nfsroot? [11:21] nope [11:21] nbd [11:21] nfs is dog slow compared to an nbd exported squashfs [11:22] amitk, try editing /lib/init/fstab and remove devtmpfs there (just keep tmpfs) [11:22] i would bet that makes it boot [11:23] though that still leave the question why parse_filesystems doesnt swee it [11:23] *see [11:25] parse_filesystems just loops over /proc/filesystems [11:25] ogra_cmpc: ok, it gets stuck at the next step now, /dev/shm [11:26] hmm [11:27] tmpfs is clearly found [11:28] though in your log it seems that all tmpfs mounts have issues too [12:25] hrw: i still intent to work on DB#577674 if you do not beat me, but not before 1-2 months [12:28] zumbi: ah... its your bug :) [12:32] hrw: sure, feel free to add (if you consider) my comments so doko is aware [12:34] doko gave mi that bug number [12:36] hrw: yes, that is current work [12:37] at least to my concern [12:37] help is very much welcome :-) [12:49] * hrw → lunch === fta_ is now known as fta [13:44] amitk: when i look at ubuntu kernel tree, i can see tags such as Ubuntu-2.6.34-1.1. what does the 1.1 mean? how does it related to mainline tag? [13:46] ndec: it doesn't relate to the mainline tag. It is Ubuntu's internal abi versioning. So 1.1 means abi version 1, upload 1. [13:46] 1.2 would be same abi but new upload [13:48] amitk: ok, thx. [13:49] ndec: this is to make sure all external modules/packages depending on internal abi are recompiled when a new kernel abi hits the archive [13:49] e.g. the dkms packages for the binary -nvidia and -ati drivers use this [13:50] amitk: so how do you decide if there is an abi change? [13:52] ndec: automated scripts (debian/script/abi-check) are run after a new kernel build. If the hashes for internal functions have changed, the build will fail if we haven't changed the abi version in debian/changelog [13:53] Has anyone used the "edit-patch" from ubuntu-dev-tools package? I'm trying to use it but it complains about "Patch system can not be detected ...." And I use quilt in my case. Any help appreciated. [13:53] amitk: thx! [13:59] berco: You could set -x it and see where it fails [13:59] berco: It might not be detecting a quilt build-dep or a quilt rules include [14:04] lool: thx. Apparently the script is looking for some file that I'm missing: "debian/patches/series" [14:05] berco: Yes, that's the expected place for your quilt series [14:06] berco: You probably want such a file with the list of patches if you're using [14:06] quilt [14:07] I have quilt installed but I'm now trying to find some doc as to what "series" is supposed to do [14:08] berco: series is a list of filenames [14:08] of patches to apply [14:08] the files should be below debian/patches [14:09] berco: You might want to look at existing packages using quilt? [14:09] berco: e.g. apt-get source gconf2 [14:10] lool: thanks. I think I got confused and thought I could use "edit-patch" to create a new patch. [14:10] berco: That's the goal I think [14:10] berco: quilt provides commands to do this as well [14:11] lool: but if I create a new patch "series" potentially doesn't exist for the 1st patch [14:11] so I need to create it empty? [14:14] Yes [14:15] lool: many thanks. "touch debian/patches/series" is enough to start working with "edit-patch" :) [14:20] berco: Cool :) [14:23] hrw: No need to put your name before work items in brackets because you're the assignee [14:24] hrw: we only do that when someone not the assignee helps with the spec [14:24] no need but I can - right? [14:25] * ogra always puts his name in front ... [14:25] ... to honor the Department Of Redundancy Department [14:28] hrw: it's a bit cluttering, but you can yes [14:32] lol :-) [14:34] hrw, there seems to be a kernel that spits out console= into upstart environment at https://edge.launchpad.net/~apw/+archive/green/+packages [14:35] seems like we'll get that for all meverick kernels soon [14:35] *maverick === rgreening_ is now known as rgreening [14:36] cool [14:38] CONFIG_INIT_PASS_ALL_PARAMS [14:39] yeah [14:39] ubuntu addon [14:39] right [14:40] yet :) [14:40] will go upstream if they want it [14:40] patch is small and quite nice [14:41] should definately be good enough to base the autotty code on [14:41] yep [14:41] waiting for omap3 kernel with it [14:41] up to amitk or cooloney i guess [14:42] or mpoirier [14:42] not sure who is supposed to take the omap3 maverick kernel [14:42] cooloney is likely busy enough with omap4 [14:43] mpoirier will take care of it [14:43] ah, right [14:45] mpoirier, hrw is working on an upstart enhancement to autospawn a getty if console=ttyS... is set [14:45] so having the patch thats being tested at https://edge.launchpad.net/~apw/+archive/green/+packages would be intresting for us in the omap3 kernel [14:46] ogra, its not like upstart cannot get that information already from /proc/cmdline [14:46] apw, Keybuk denied that [14:46] apw: keybuk does not want that way [14:46] he wants to have the script proper on first shot :) [14:46] already discussed and it got refused [14:46] i can believe he doesn't want it that way, but till he tests the patch for real its in it court [14:47] i suspect there is some initramfs work to stop that fecking up all the options too [14:50] so, should I try applying the patch on OMAP3 ? [14:50] mpoirier, as soon as apw approves and Keybuk has tested it [14:51] very well. Lucid, Maverick or both ? [14:51] (just wanted to let you know why i pinged you actually :) ) [14:51] maverick only [14:51] should we open a bug for it ? [14:51] lucid doesnt get new features [14:52] mpoirier: did not omap3 built from same sources in maverick? [14:52] indeed. [14:56] hrw, well, a bug wont do any harm to track the issue [14:56] so you actually know when its applied [14:56] hrw, feel free to subscribe to bug 586386 [14:56] Launchpad bug 586386 in linux-ti-omap (Ubuntu) "omap3 kernel should hand over all comdline args to the init environment (affects: 1)" [Undecided,New] https://launchpad.net/bugs/586386 [14:56] bah, i typoed :( [14:57] ogra: let apw know about the bug so that he can auto close it when he applies it to the kernel [14:58] indeed [14:58] apw, ^^^ [15:01] added links [15:05] ogra, amitk, ack have the number now [15:05] thanks ... i would have triaged it properly but dont know what that takes for the kernel team :) [15:55] have you guys seen meego release? [15:55] looks like moblin [15:56] it is moblin [15:56] is it done in Qt? [15:56] just rebranded [15:57] amitk: no, still clutter/gtk [15:57] or rather MX (which is clutter based UI toolkit) [15:57] hmm [15:57] they use qt as the "standard" toolkit though [15:58] zyga: they announce it atleast [15:58] * hrw had to work in moblin team [15:58] there is some talk about how to use qtcreator to make meego apps [15:58] and since meego = nokia + intel [15:58] qt and atom are the way to go for this project [15:58] I wonder how it plays out with ARM and ... AMD [15:59] meego is this nice "not android" software you can put on a product [16:00] * zyga just tried dd'ing meego image on to /dev/sda on his main workstation [16:00] fortunately I was saved by crappy gvfs [16:00] the image was on .gvfs [16:01] and sudo dd didn't find the source file [16:01] this made me re-read what I wrote [16:02] that error made me write usb-imagewriter :) [16:02] * ogra once wiped his laptop HDD with a wrong dd [16:04] I have system on sdb [16:04] sda1 is 10GB very old debian rootfs so no big loss (newer is on sdb1) [16:05] well, if you typo your dd command to sdb thats bad then :) [16:06] ogra, the one that ubuntu has? [16:07] ogra: current rootfs is on sdb3 [16:07] zyga, ? [16:07] the one what ? [16:07] ogra, what is usb-imagewriter [16:07] its a gui for dd [16:07] its in universe [16:08] and scans for mounted disks before offering you devices to write to [16:08] so you can never write to your actual root disk [16:08] * zyga will try it out [16:09] file bugs if you find any, i havent used it since quite a while :) [16:13] lool: the more I look at dh_movefiles removal, the more I do not see a need for it. [16:13] lool: I can do migration (already migrated some subpackages) but it can be ugly in few places. [16:47] ogra - you around ? [16:48] mpoirier, yup [16:48] good. [16:49] bug 566645 [16:49] Launchpad bug 566645 in linux-ti-omap (Ubuntu Lucid) (and 1 other project) "doing a netinstall to SD card results in OTG related oops on first boot (affects: 2) (heat: 14)" [High,Confirmed] https://launchpad.net/bugs/566645 [16:49] have you been able to reproduce reliably ? [16:50] not since OTG was disabled completely anymore :) [16:50] educate me on OTG pls. [16:50] its the small USB port on the C series beagle [16:51] it can run in host or gadget mode [16:51] ok. [16:51] what as OTG disabled ? [16:51] during lucid development we found its broken, amitk didnt manage to find a fix so it was disabled completely [16:51] when was OTG disabled ? [16:51] after i filed that bug [16:51] we indeed want it to work in maverick [16:51] is the bug still valid ? [16:52] dunno, once i have a maverick image to test i will be able to tell you, its not valid for lucid anymore [16:52] vey well. [16:53] i changed the bug status accordingly [16:53] about maverick. [16:54] I have a lucid powered beagleboard. [16:54] I compile a 2.6.34 kernel and install it on. [16:54] ok [16:54] have you done it ? [16:55] /dev/mmcblk02 seems to be missing, preventing the rootfs to be mounted. [16:55] and the system stalls in busybox. [16:55] I'd like to know if I'm the only one with the problem. [16:55] /dev/mmcblk02 or /dev/mmcblk0p2 ? [16:55] the latter yes, [16:56] writing this off the top of my head. [16:56] i havent put much time into omap3 stuff recently working on the omap4 stuff atm [16:56] ok, I'll investigate [16:56] but i just compiled a touchbook kernel for my touchbook which works fine [16:56] based on .34 [16:57] so its likely a config option you miss or something like that [16:57] since the touchbook boots from mmc [16:57] Well, it's off Leann's integration branch [16:58] which should include all the OMAP3 changes from Lucid. [16:58] hrw: problem is that dh_movefiles is deprecated [16:58] hrw: dh compat 2 is really old and might go away [16:59] ok, but what does it give except 'dh_movefiles may be dropped one day'? [16:59] hrw: That seems like a good reason to move away from it [16:59] ogra: you already have a panda board don't you ? [16:59] mpoirier, if we have a binary kernel i'll help testing that on my beagle its hard to tell why yours doesnt find mmcblk0p2 [16:59] hrw: but frankly, it's not the highest priority work in the cross-compilers spec [16:59] mpoirier, not yet, it should arrive on the weekend or monday i think [17:00] mpoirier, did yours arrive already ? [17:00] ogra: nop, but was supposed to have been shipped. [17:00] lool: sure, we need to prioritize it [17:00] great [17:00] ogra: what are you testing OMPAP4 on then ? [17:00] blaze [17:00] lool: but this updated my debhelper knowledge [17:00] looks like a giant mobile phone :) [17:01] hrw: Ok [17:01] hrw: Are you still fighting it [17:01] lool: I made some notes and can return to it later without problems [17:01] ogra: for mmcblk0p2, all you need is the linux-image*.deb file ? [17:02] hrw: Ok, if you're moving to something else, mark it back TODO [17:02] mpoirier, did you do a standard ubuntu install on the board (i.e. from one of the installer images) [17:02] sure lool [17:02] then you should just be able to dpkg -i the deb [17:03] ogra: I'm in Lucid and do "dpkg --force-depends -i package.deb" [17:03] why --force-depends ? [17:03] ogra: --force-depens is mandatory. Otherwise complains about wireless package not present on system [17:03] then you didnt do an ubuntu install :) [17:04] the images ship these packages by default [17:04] so you will additionally need to create an uImage from vmlinuz as well as a uInitrd [17:04] ogra: I followed the instructions found at http://elinux.org/BeagleBoardUbuntu#U-Boot_uImage [17:04] ow [17:04] ? [17:05] well, while thats nice to get ubuntu running its not ubuntu :) [17:05] our setup is different, the installer applies bits these images cant provide (like writing uImage to NAND etc) [17:05] so if you use such an image you have to do some parts manually to upgrade the kernel [17:06] if you use an ubuntu kernel package [17:06] first of all i'd resolve the dependency issue so you dont have to use --force-depends all the time :) [17:06] after dpkg -i failed run "sudo apt-get -f install" [17:06] that will install the missing package [17:07] before going any further it is mandatory that I get my board configured the same way as yours. [17:07] (assuming you have a NIC attached) [17:07] I do. [17:07] nah, not for kernel testing [17:07] actually not using NAND for the kernel will make testing easier [17:08] but you should use a script for the updating of the kernel ... [17:08] http://paste.ubuntu.com/439253/ thats a setup i use on my blaze [17:08] not using NAND is a boot script thing. [17:08] that should update the kernel on vaft [17:08] *vfat [17:08] mpoirier: cheap omap boards after beagleboard will not have nand afaik [17:09] * armin76 steals NCommander's boards [17:09] hrw, right, but flash-kernel has no integration for that setup yet, i still need to write it [17:09] it will be similar to the above paste [17:10] just less hardcoded :) [17:11] ah.. no symlinks on vfat.. [17:11] but do we have symlinks in /boot? [17:11] and using a vfat bootpartition is surely better than having to fiddle with uImage in nand from the u-boot prompt ... if you run test kernels taht break from time to time [17:11] we do [17:11] omap has.. [17:11] which is why we will have an extra vfat partition for the uImage [17:12] my desktop does not [17:12] right [17:12] ogra: on bug 2.0 we used one ext3 partition and enabled ext support in uboot [17:12] worked fine [17:12] bug 2.0 is omap3 based [17:13] kernel was loaded from /boot/ on that ext3 [17:13] well, on beagle u-boot ext2 support always crashed for me or took horridly long [17:13] before we used same config on i.mx31 [17:13] anyway time for me [17:13] ogra: get back to me when you're done with hrw [17:13] right, if ext support works ext is great [17:13] have a nice rest of day [17:14] ciao hrw === hrw is now known as hrw|gone [17:14] mpoirier, just chime in that was just chatter :) [17:14] ogra: greetings [17:14] ogra: hey, is there a trick to using an unbuntu rootfs over NFS? [17:14] mpoirier, so use something similar to the script i pasted above and put it into kernel-img.conf [17:15] prpplague, amitk did that today shouldnt require any tricks [17:15] ogra: hmm [17:15] mpoirier, as a postinst_hook [17:16] ogra: we tested on panda and blaze, and both hang after mounting the rootfs and starting up the GUI stuff [17:16] mpoirier, the dpkg -i usually rebuilds the initramfs if you use ubuntu kernel packages so it should just work [17:16] prpplague, but you can boot it up to a console prompt ? [17:17] ogra: everytime I updated it did rebuild the initrd [17:17] might be an issue with gnome [17:17] mpoirier, perfect so the above script should just work for you [17:17] ogra: I need to understand the kernel-img.conf thing. [17:18] its a file usually created by the installer on arm we dont actually use it anymore [17:18] ogra: I was already doing mkimage to generate a new uImage and uInitrd - classic embedded stuff [17:19] right, the postinst script of the kernel package looks in kernel-img.conf, if it finds a postinst_hook entry it will execute it after the package was installed [17:20] we used to use flash-kernel as a postinst_hook in former releases (flash-kernel is the tool used on all arm platforms in debian and ubuntu to care for HW specific cases to activate the new kernel ... i.e. write to NAND if needed or copy to vfat etc) [17:21] today flash-kernel is automatically called by update-initramfs so kernel-img.conf isnt used anymore [17:22] as i said above, the changes to flash-kernel to create an uImage on a vfat partition were not written yet so currently you need to apply a script and use kernel-img.conf for your setup [17:22] ogra: yea if we use a scaled down image we can get a prompt [17:22] i'll land these changes soon (before the alpha2 release) [17:23] ogra: but if we use the larger images it just hangs [17:23] prpplague, very likely gnome not getting along with NFS mounted home [17:23] do you use an initramfs ? [17:24] (that uses nfsmount and applies nolock in ubuntu, i'm not sure the kernel does that by default if you use the in-kernel way of nfsroot) [17:24] we weren't using an initramfs for this testing, i was guessing it probably had some permissions or temp workspace issues [17:25] ogra: ahh seems like i read something about that with regards to ubuntu [17:25] well, if it boots through to a running system its likely X related ... gdm or the homedir in gnome etc [17:25] ogra: is there a wiki page or howto doc we could look at? [17:25] for using initramfs ? [17:25] or for nfs bugs ? :) [17:25] for doing ubuntu over nfs [17:26] i dont think there is anything arm specifc [17:27] https://help.ubuntu.com/community/DisklessUbuntuHowto might be the best you can get but thats outdated i think [17:27] thanks i think i had that one bookmarked for a look over today [17:27] ogra: I looked at you pasted script and I was doing the exact same thing, except mine wasn't automated. [17:28] ogra: the real problem seems to come from the initial installation [17:28] that might be [17:28] ogra: where some packages seems to be missing. [17:28] ogra: I don't think it is cause by my "C2" rev. [17:28] right, that rootfs is created with rootstock that might not get you a 100% ubuntu [17:29] its shouldnt be [17:29] *it [17:29] ogra: that being said, where do I find the "approved" way of installing Ubuntu on beableBoard ? Should I use the netinstall ? [17:29] prpplague, sorry that i cant help much more but since i switched ltsp to nbd images in 2006 i havent touched nfsroot anymore :) [17:30] hehe indeed, it was just one of the items that we wanted to test [17:30] ahh the good ol' days of LTSP [17:30] mpoirier, you wont have much fun for kernel development with the approved way [17:31] mpoirier, since we flash the kernel to NAND by default which gets you into bad situations once you flashed a broken kernel [17:31] mpoirier, though https://wiki.ubuntu.com/ARM/Beagle has the 2official" images [17:33] ogra: haven't i seen you in the #ltsp channel in the past? [17:34] prpplague, i wrote ltsp5 with scottie, you have for sure :) [17:34] i stepped down in 2008 though [17:34] ogra: ahh right [17:34] * prpplague dropped out of LTSP stuff back around 2001 [17:35] but i'm still a resident there and go to the november meetings in maine :) [17:42] btw, semi-related to questions about X11/gdm and nfsroot, etc... how actually does recovery mode work? [17:43] is there some bootarg I can add to tell upstart not to start gdm/x11? [17:44] single or text on the cmdline [17:44] ogra: ahh, perfect.. thx [17:44] single requires an initramfs and gets you to a rootshell, text just switches off gdm [17:44] ok.. text is probably good for what I am trying to do.. === sbambrough is now known as sbambrough-lunch [18:24] ogra: the package I'm missing is "wireless-crda" [18:25] yeah [18:25] any idea why and where I could get it - for ARM that is. [18:25] just apt-get -f install should install it [18:26] beyond that http://ports.ubuntu.com/ubuntu-ports/pool/main/w/wireless-crda/ but apt knows which version you want [18:26] fabulous - thanks. [18:46] prpplague: ogra: I didn [18:46] oops [18:47] prpplague: ogra: I didn't use the full desktop, only build-essential and openssh-server for my rootstrap [19:51] omapfb omapfb: illegal display bpp [19:51] omapfb omapfb: failed to setup fb_info [19:51] omapfb omapfb: failed to setup omapfb [19:51]