=== Baybal__ is now known as Baybal === XorA|gone is now known as XorA [10:26] ogra_cmpc: nice [10:26] (27M for X) [10:43] asac, do you mean RAM ? [10:46] yes [10:46] ogra: thats what you said, right? [10:47] thats the whole system [10:47] ogra: what plymouth theme to use for a console only image? [10:47] or can i get rid of that completely? [10:47] when i remove the initramfs-tools scripts all is fine ... if not i stay on a console 7 with a SEGV of plymouth and have to manually go back to the console 1 etc. [10:47] mountall needs plymouth [10:47] i know [10:47] but not the initramfs hooks ;) [10:48] (it seems) [10:48] there was a command to select the theme [10:48] * ogra forgot which [10:48] ogra: but which theme to use? [10:48] text [10:48] plymouth-theme-text -> SEGV [10:48] ubuntu-text works i think [10:49] plymouth-theme-ubuntu-text [10:49] is that installed ? [10:49] /lib/plymouth/themes/ubuntu-text/ubuntu-text.plymouth [10:50] i will try that again now ... but plymouth-theme-text crashes for sure [10:50] its what i get on my server installs [10:50] ogra: i made an autologin by hacking tty1.conf ... is that the "normal" way? ;) [10:51] for a text console ? [10:51] yes [10:51] its usrely not "normal" but likely a good way to do it :) [10:51] i made a /usr/bin/autologin script that just run login -f root [10:51] *surely [10:51] and it works. [10:51] just wastn sure if thats good [10:51] we had something for that in the ume images [10:52] bceause tty1.conf you cannot really change from outside .... dpkg-divert on a conffile feels painful [10:52] which was a bit cleaner [10:52] ogra: whats the clean way? [10:52] an upstart job [10:52] assume there is no user installed ;) [10:52] i need to dig up the old code [10:52] hmm [10:52] well, why isnt there any user ? [10:52] tty1.conf is an upstream job [10:52] create one during build [10:52] you can create a fresh upstart job [10:53] no need to mangle an existing one [10:53] hmm. but tty1 is taken ... or can i just take any tty? [10:53] how can i ensure that that tty is the default after boot? [10:53] no idea :) ask a plymouth dev :) [10:53] heh [10:54] you should be able to use console= on the cmdline but that will kill plymouth [10:54] yeah [10:54] let me first try if plymouth is happier with tubuntu-theme text installed [10:55] in the meantime i have to cleanup this disk ... not enough space to copy a live netbook image to it [10:55] http://bazaar.launchpad.net/~ubuntu-mobile/ubuntu-mobile/ubuntu-mid-default-settings/files [10:55] that should have the old upstart files [10:56] ok thanks. will check that out [10:56] (if my browser ever opens) ;) [10:56] x [10:57] that stars the gui ;) [10:57] http://bazaar.launchpad.net/~ubuntu-mobile/ubuntu-mobile/ubuntu-mid-default-settings/annotate/head:/session [10:57] starts [10:57] modify that [10:57] and use it as upstart job [10:57] yes, it starts the gui [10:57] so you have to modify ;) [10:58] right. so you dont hvae to bother about switching to a "new" tty [10:58] righ, but *you* can use openvt and switch to tty9 or so ;) [10:58] iirc openvt just takes the next free tty [10:59] hmm. there is a chvt command [10:59] chvt needs a number [10:59] ogra: putting that switching feels like a racy thing. [10:59] openvt checks for the next free one or so [11:00] yeah [11:00] is that a replace for getty? [11:00] i think not [11:00] you still need to run login or something [11:00] try it :) [11:01] i havent touched that stuff since 1.5y [11:01] yeah [11:01] will try ;) [11:04] my tries "on started tty1 -> chvt 1" ;) [11:05] that wont work [11:05] since you already have a getty running there [11:05] hmm. but its chvt [11:05] you will end up with two processes controlling the tty [11:05] hmm [11:05] which will get you funny output if you type anything :) [11:06] its just chvt [11:06] it manpage it says its similar to using f1 etc. [11:06] sometimes one sometimes the other process will reply to enter [11:06] oh, indeed, if you dont run login it will just change the vt [11:06] hmm. but chvt isnt really controlling anything [11:06] i thougth you wanted autologin too [11:06] yes, thats the idea [11:07] ogra: oh right ;) [11:07] yeah [11:07] i still have the hacked tty1.conf so it auto logs in [11:07] so you need a fresh tty [11:07] use openvt [11:07] if thats for a higly embedded system you should cut down the amount of ttys anyway though [11:08] i think each running getty will eat up to 800k [11:08] ogra: how can i get the ttys down ... they are all shipped by upstart :( [11:09] at least without forking [11:09] * ogra would recommend an upstart-embedded package that has all that stuff adjusted [11:09] build from the same source but ship other configs and conflict with upstart [11:09] or some such [11:10] seems cleaner than diverting config files [11:11] yeah [11:11] either that or use openvt :) [11:11] ok kbd isnt installed in an essential only system [11:11] * asac adds that [11:12] yeah [11:12] gpgv also [11:12] ogra: i have added gnupg [11:12] whta do you need gpgv for? [11:12] well, gpgv is enough [11:13] apt needs it for the Release.gpg [11:13] right. i need gnupg for apt-key [11:13] to make the ppa fly [11:13] ah [11:13] yeah, that should bring in gpgv anyway [11:13] good ;) [11:14] you also want to add something to /etc/apt/apt.conf.d/ that makes not installing recommends the default [11:14] ogra: so where is the omap uboot? [11:14] didnt we package that? [11:14] yes [11:14] it uses the upstream naming [11:14] ogra: i am not installing recomments [11:14] nds [11:14] look for u-boot [11:14] ah [11:15] i have to switch the other u-boot packages too or revert that in maverick [11:15] not sure yet [11:15] there it is ;) [11:15] ogra: so what do we chip for omap in lucid? [11:15] do we put uboot on it at all? or assume its all setup on beagel? [11:17] in lucid we assume its in NAND [11:17] for maverick i have to come up with something clever [11:18] since some of the future HW wont have NAND [11:18] for beagle we dont need to care though ... there is NAND for all released versions [11:20] ogra: where does the uboot we assume is there look for boot.scr? [11:20] on fat in / ? [11:20] iirc on fat on mmc0 0:0 [11:23] kk [11:27] asac, ./include/configs/omap3_beagle.h in the u-boot-omap3 source [11:28] specifically CONFIG_EXTRA_ENV_SETTINGS and CONFIG_BOOTCOMMAND [11:30] hmm. guess that uboot cant do ext2load? [11:30] seems it can :/ [11:30] not on omap [11:31] fails miserably [11:31] too bac [11:31] d [11:31] yes [11:31] especially soice you cant put /boot on fat [11:31] *since [11:31] cant? [11:31] nope [11:31] whats that? [11:31] dpkg uses hardlinks when it unpacks files [11:32] hmm. but we use fat on all images for /boot, dont we? [11:32] so it creates a vmlinuz-$(uname -r).bak when it tries to put the new file in place ... that file is a hardlink [11:32] no [11:34] one thing i was trying is to have a separate fat patrition that flash-kernel writes uImage, uInitrd and boot.scr to but for lucid lool convinced me to rather use NAND [11:34] in maveric where we cant use NAND for everything i think i'll have to revisit that setup [11:34] +k [11:35] ogra: how does the sdcard boot then? not oob? [11:35] the images are on vfat so there u-boot can read boot.scr [11:36] during install i write the boot options, kernel and initrd to NAND [11:36] so then it doesnt matter [11:36] it only matters on systems that have no NAND but we dotn support any with the lucid install [11:36] hmm [11:36] maverick *will* support these though [11:37] which means we'll need a separate fat somewhere [11:37] we need it for this directly booting rootfs setup too [11:37] right, that needs tweaks in flash-kernel [11:38] so once linux-image is installed it needs to temporary mount the fat and write uImage and uInitrd to it [11:38] (or after update.initramfs) [11:38] ok at least plymouth ubuntu theme text doesnt play games on me anymore ;) [11:38] heh [11:39] now back to openvt ;) [11:42] ogra: does uImage in nand win over uImage on fat mmc0? [11:42] see the config i pointed to above :) [11:42] e.g. could we live with shipping "not changing" uImages on some fat partition? and flash-kernel updates the nand [11:43] it only cares for boot.scr [11:43] err, no [11:43] actually it doesnt [11:43] http://paste.ubuntu.com/426935/ [11:44] asac, how would you update a nonexisting NAND ? :) [11:44] i will use NAND on all boards where we know we have it in flash-kernel [11:45] well. i am currently targetting beagle. .. but without installer [11:45] flash-kernel needs to detect the hardware anyway and needs special casing for each board [11:45] without installer, hmm [11:45] then you need MLO and u-boot.bin as well as boot.scr [11:46] and a fat partition [11:46] loadbootscript=fatload mmc 0 ${loadaddr} boot.scr [11:46] so you also need the first partition being the fat [11:49] though note that if you use flash-kernel from lucid it will automatically write to NAND atm [11:49] and assumes your system is properly set up for uboot-envtools [11:50] (which is done by flash-kernel-installer) [11:51] http://paste.ubuntu.com/426937/ [11:51] ogra: yes, but for that to work we would need to prefer nand over fat as fat will become outdated ... which creates a chicken egg problem etc. [11:51] (thats what flash-kernel-installer does at the end of d-i/ubiquity [11:51] so plymouth is still broken even with ubuntu theme ... was tricked because i still had my hand hacked image loop mounted and umounting it seems to have wiped the new img [11:52] (not that it wipes the u-boot defaults so it wont look for boot.scr anymore) [11:52] ah [11:53] so it wiped the defaults [11:53] good [11:53] only if you used the ubuntu install [11:53] that would work. e.g. put just the uboot bits on a first fat partition [11:53] once flash-kernel runs all will be nand [11:53] right [11:53] but you somehow need to run that flash-kernel-installer.postinst bit [11:54] which is tricky without any kind of installer [11:55] flash-kernel-installer is a udeb and not installed anywhere but in d-i/ubiquity [11:58] ogra: why isnt cmdline="root=UUID=$uuid ro quiet splash vram=12M omapfb.mode=dvi:1280x720MR-16@60 fixrtc" [11:58] etc .that run in normal flash-kernel? [11:58] e.g. the stuff from http://paste.ubuntu.com/426937/ [11:59] flash-kernel (like update-grub) doesnt touch the cmdline [11:59] creating the cmdline is job of the installer [11:59] after install that shouldnt be changed by any bootloader tool [12:00] unless yuo change it manually (by editing /etc/default/grub or in case of beagle using fw_setenv) [12:01] flash-kernel only flashes the kernel (and initramfs) [12:01] flash-kernel-installer sets up the environment for flash-kernel during installation [12:03] http://paste.ubuntu.com/426943/ <- flash-kernel ... http://paste.ubuntu.com/426937/ <- flash-kernel-installer [12:04] * ogra takes a break [12:05] ogra: so openvt didnt work ... chvt worked fine ;) [12:05] probably just working around the fact that plymouth still fails miserably [12:10] how did you call openvt ? [12:10] sure [12:10] sure ? [12:11] yes. it opened a new terminal on all other terminals [12:11] so i got funny stuff like login -> hit return -> root shell -> hit return -> login prompt [12:11] etc. [12:12] what was yous commend ? [12:12] *your [12:12] * asac tries again [12:12] exec /bin/openvt /bin/autologinroot [12:12] you need at least -s [12:13] * asac checks manpage [12:13] probably even -e [12:14] /bin/openvt -e -s -- /bin/autologinroot [12:14] or [12:14] /bin/openvt -e -s -- su -l root -c /bin/bash [12:15] just copy the original tty1.conf for that and replace the exec [12:16] well, after exec, exec needs to persist [12:18] ok that worked ;) [12:18] :) [12:19] a bit dumb that its now on tty8 [12:19] but well. so be it [12:19] plymouth is still unhappy, but thats normal it seems [12:21] for flash-kernel-installer you could do some ugly first login magic in /etc/bash.bashrc [12:21] look for "if [ ! -e "$HOME/.sudo_as_admin_successful" ]; then" in there [12:22] though thats not very sexy [12:22] great. yeah [12:23] grr. something is editing 10-help-text :( [12:23] in update-motd.d [12:23] i get a conflict when i change that ... dont get that for other conffiles i change in the same way [12:24] 10-help-text ?? where does that belong to ? [12:24] base-files of course [12:24] or where doe it live [12:24] like verything related to branding is stuffed into the corest of core things ;) [12:24] /etc/update-motd.d [12:24] hmm [12:24] ah [12:25] and seems something is messing with that during a bootstrap [12:25] e.g. changing it post install [12:25] update-motd likely [12:25] what does update-motd do with it? [12:25] it shoud just source it [12:26] guess i will sed it in postinst ... seems its modified anyway, so nothing to loose [12:26] hmm, indeed [12:28] the conffile mechanism is really a mess in ubuntu/debian ... folks seem to not get it. you need a) package defaults + b) admin defaults (thats conffiles) and user changes in $HOME ... a) and b) are far too often not separated, but just shipped in etc [12:29] well, ucf will only kick in if there is a three way diff [12:30] i.e. if two defaults changed at the same time [12:32] what is ucf? [12:32] update config file [12:32] the tool that prompts during debconf [12:32] ah [12:33] yeah. well. the fact that there were changes int he first place scares me [12:33] i hope its some hack cody added to the live-helper i didnt find yet [12:33] ok /me spins new image and hopes for the best [12:41] live-helper surely shouldnt touch it [12:41] else thats a bug [15:19] ogra, Do we have support for ethernet gadget in latest 2.6.32.11-l13 ?? === XorA is now known as XorA|gone === hrw|gone is now known as hrw === hrw is now known as hrw|gone === jmcgee is now known as jmcgee|gone