=== gry is now known as gry2 === gry2 is now known as gry [04:47] i was going to suggest he add rootwait to his bootargs [04:47] some mmc devices are rather slow [09:31] I get this while booting kernel 3.3.1: "EXT3-fs (mmcblk0p2): error: couldn't mount because of unsupported optional features (240)" [09:31] then "EXT4-fs (mmcblk0p2): warning: maximal mount count reached, running e2fsck is recommended" [09:31] and then "EXT4-fs (mmcblk0p2): mounted filesystem with ordered data mode. Opts: (null)" [09:32] and last "VFS: Mounted root (ext4 filesystem) on device 179:2." [09:32] the partition is a ext2 partition, but still it seems ext4 mounts it [09:33] the last line in the log is "Freeing init memory: 248K", but I still don't get any login console [09:33] should I? [09:34] you sure the recomended fsck isnt running? [09:35] no. how do I know? [09:36] no idea unless you have disk activity light [09:36] also not sure which ubuntu versions give a console prompt and which just go to desktop with nothing on serial [09:37] I don't think I do. It's a small ARM system (Cubox). But the partition is only 4 GB, so I figured it would be done by now if so [09:37] I'm running Ubuntu Core 12.04 beta2, there's no desktop to go to [09:38] you could try fscking the card on a PC [09:38] it may have got stuck with an unrecoverable error or something [09:38] but thats all I can think off [09:38] good idea. Would it be enough to just reformat the partition? [09:39] I always try and recover before getting drastic [09:39] just to try and understand what happened [09:39] ok, I'll try that [09:44] it found some errors, fixed them, recreated the lost+found and the error message "EXT4-fs (mmcblk0p2): warning: maximal mount count reached, running e2fsck is recommended" disappeared [09:44] so it did something :) [09:44] still it cannot mount the filesystem using ext2, even if it is an ext2 filesystem. it mounts it using ext4 [09:44] and still no login console [09:44] at some point in time you mounted it as ext4 and wrote a file [09:44] hmm [09:45] perhaps if ubuntu did it automatically when I inserted the SD card in another computer? [09:45] possibly [09:45] but as to why its not booting no idea [09:46] ubuntu doesnt do such stuff, does your kernel have ext2 support ? [09:46] yes, its compiled into the kernel [09:46] if not, it will indeed try different FS drivers [09:46] also does your fstab have a proper entry for it pointing to ext2 [09:48] no, it does not. that file is just empty [09:48] (heh) [09:48] well, that means you leave it to the kernel to do guesswork :) [09:49] (ubuntu-core is seriously not thought as a rootfs if you dont know how to configure it and do all bits the installer would do ...) [09:49] I am having a freezing problem :/ after "init: unreadahead main process (105) terminated with status 5" it hardlocks [09:50] are you sure it hardlocks or does it just not display anything ? [09:50] well, I am learning. I had Ubuntu Core running on qemu arm, and the only modification I did to get a login prompt was to setup a root password there. [09:50] yeah it hardlocks [09:50] weird [09:50] ceck your boot.log ;) [09:50] *check [09:50] I have an image on sdcard that should just go to console login rather than starting x [09:50] and also logs are not written to [09:51] you dont have the SD locked to readonly, do you ? [09:51] this is both emmc and microsd [09:51] hmm [09:51] on transformer prime [09:51] we think it is probably the kernel, just don't know why, and seriously lacking some debugging options [09:51] your kernel has all necessary bits ? [09:51] yes [09:54] http://paste.ubuntu.com/939433/ <-- my config [09:54] just incase you do see something I may have missed [09:54] bug 936667 came to mind, but you should at least see error messages [09:54] Launchpad bug 936667 in upstart "Upstart early job logging causes boot failure for systems with no initramfs (error is "No available ptys")" [High,Fix released] https://launchpad.net/bugs/936667 [09:55] well I do have a different error when running noinitrd [09:55] but it still ends up in the same lockup [09:56] hmm, you have legacy pyts on [09:56] not sure if that can get in your way, but the latest upstart is picky about kernel options at times [09:56] oh that was me trying to work around failed to create pty with noinitrd [09:56] I had this problem without legacy ptys [09:57] wow, why do you compile loopdevice support into the kernel ? [09:57] 8instead of making it modular) [09:58] cause that is how it is for android [09:58] I haven't finished with the kernel yet [09:58] wow, and also DM [09:58] your kernel must be huge [09:58] 3.5MB [09:59] this is just how things are in the default android config that I haven [10:00] ogra_: do you know any place I can find the minimum configuration I need to do in order to get ubuntu core up and running? [10:00] haven't bothered changing yet [10:00] wowhmm your wlan driver is also hard compiled in and has a hardcoded firmware part ... [10:00] yeah, that is bcmdhd [10:01] jeinor, http://omapedia.org/wiki/OMAP_Ubuntu_Core has some hints [10:01] thanks [10:01] the firmware part is hardcoded cause I needed to rearrange some things :) [10:01] lilstevie, i would make a lot stuff in your kernel modular [10:02] as a first step [10:02] bcmdhd does not work, at least on android without being compiled in [10:02] this is a very early 3.1.10 port, a lot of the asus stuff is still horribly intertwined [10:04] well, if you are sure it is locked up hard ... hard to say what it is ... if you wouldnt be sure about that i would just have guessed for a worng console= arg on the cmdline [10:05] the commandline is certainly correct [10:05] (the ureadahead message is usually the last bit you see before the userspace console kicks in) [10:05] and this kernel will boot android with its config differences too [10:06] and yeah I know [10:06] well, that it boots android has nothing to say :) [10:06] thats why I was wondering if there was something extra for a more verbose output [10:07] and haha yeah .39 boots android too, but has a horribly broken framebuffer in ubuntu [10:07] since you seem to properly get through the initrd i would boot with something like break=top or break=premount or so in your cmdline [10:07] that will drop you into the initrd shell for further debugging [10:07] ok [10:08] (and indeed make sure to not have "quiet" set ;) ) [10:09] jeinor, note that the above url refers to pandaboards, so dont pick up things like bootloader4 config etc from there since this is HW specific [10:10] yeah I only have root= at the moment [10:10] ogra_: got it. I used it for configuring network and root user so far [10:11] I also recreated the filesystem using mkfs.ext3 and reuntared ubuntu core to it [10:11] let's see what happens... :) [10:11] lilstevie, right, try break=top ... that should get you the initrd prompt right after uncompressing [10:14] ogra_: that removed all error messages on the filesystem. it now properly detects the filesystem as ext3 and mounts it without any error messages. [10:14] good [10:15] ogra_: i still get no login console, but I'm starting to believe there's a problem with the graphics driver. I see no output whatsoever on the monitor connected via HDMI during the boot process [10:15] well, do you see anything on the serial console ? [10:16] I don't think I have setup a serial console. Learned yesterday how to connect through putty to be able to view the output to ttyS0 [10:16] so I see all kernel messages in the putty console window [10:16] but I don't think I can do anything from there [10:17] well, see the hints on the webpage for setting up serial login then [10:17] right [10:25] ogra_: testing with break=top now [10:25] good [10:25] ogra_: how do I connect to my new serial console? do I use putty? [10:25] do you use windows on your host PC ? [10:25] no [10:25] ubuntu [10:26] why do you use putty then and not a sane terminal program ? [10:26] putty is a windows terminal emulator (that happens to have been ported to linux) [10:26] the guide from the Cubox wiki contained instructions for using putty [10:26] I see [10:27] what is a sane terminal program? [10:27] try screen ... [10:27] i.e.: [10:27] screen /dev/ttyUSB0 115200 [10:27] that should get you a serial console in the terminal you run it in [10:28] (see man screen for the commands to attach/detach from that session etc) [10:28] that gave me a flashing prompt. Would I need to switch tty using ctrl+alt+fX? [10:28] no [10:29] ok. that SHOULD have given me a login prompt then, right? [10:29] it talks directly to your serial port and will exactly output what the cubox sends to the serial port [10:29] (you should see boot messages (and possible errors) if you reset the cubox) [10:31] yep, that gives me the same output as putty did before (but nicer in a normal terminal window) [10:31] thanks for the tip [10:31] interesting [10:32] ogra_: it boots when exiting from the initramfs shell [10:32] but if I had succeeded in configuring the serial-console, I should have gotten a login prompt here, correct? [10:32] lilstevie, race then ... [10:32] heh fun [10:32] jeinor, not if there were errors during boot [10:32] ogra_: what do you suggest [10:32] jeinor, reset the cubox and see if you see any boot errors now [10:32] lilstevie, debugging :P [10:33] heh [10:33] try break=premount next ... that drops you into the console a bit later [10:33] if that works too, try =bottom ... that gives you the console after it mounted the rootfs [10:34] ogra_: same output as before, ends with "Mounted root (ext3 filesystem) on device 179:2", "devtmpfs: mounted" and last "Freeing init memory: 248K" [10:34] ok [10:34] I'll check earlier in the log [10:34] jeinor, whats your kernel commandline ? [10:35] i.e. what does the bootloader tell the kernel when it loads it [10:35] "console=tty0 console=ttyS0,115200n8 vmalloc=384M root=/dev/mmcblk0p2 rootdelay=2 clcd.lcd0_enable=1 clcd.lcd1_enable=0 --no-log --verbose" [10:35] are you sure you need the clcd bits in there ? [10:36] no, I'm not sure what they do. I copied them from the stock kernel command line [10:36] Tried to google that, didn't find anything so I figured I'd better leave them as is [10:36] well, i guess the device has two lcd ports on borad [10:37] LCD = HDMI/DVI/VGA? [10:37] no idea how marvell defined them :) [10:37] ok :) I'll try removing them [10:38] one could be attached to the hdmi port and one to an internal socket that gets used in other devices for touchscreens etc [10:38] makes sense, there is one HDMI port on the Cubox, no VGA/DVI [10:40] btw, are you using the armel or armhf tarball ? [10:41] currently the armel [10:41] eek, [10:41] k [10:41] not using to your full potential! [10:41] that should be on the safe side then [10:41] not using at any potential right now... :) [10:41] jeinor, you know you can always use multiarch to install armel packages if you have problems right? [10:41] right, but it will make sure he wont be hit by any floating point issues for now [10:42] getting a first boot up is the current focus ;) [10:42] ogra_, kernel could still be compiled with VFP support, but that kernel floating point but a few revs back only affected x86-32 [10:42] *bug [10:42] the kernel doesnt care [10:43] ogra_, it actually has a VFP option [10:43] scientes: didn't know that. Will try the armhf as soon as I got an armel up and running [10:43] (wouldn't make much of a speed diff as there is *very* little fp in the kernel) [10:45] how do I scroll in screen? the --verbose option gives lots of output [10:45] jeinor, pipe it to less? [10:45] | less [10:45] man screen (in a second terminal you put next to it) [10:45] scientes, wont help much with dmesg from a serial port :) [10:46] ogra_, well if you have that you can just turn on infinite back scrolling in your terminal emulator [10:48] uhm... feeling a bit stupid here... what does "C-a" mean? what key is that? [10:48] C means control [10:49] a means a :) [10:52] ^ [10:52] :) thanks. entered copy mode, and now I can scroll in the buffer [10:52] init is doing a whole lot of work here [10:53] but there are no error messages, just a lot of "Handling start event" [10:54] I must have done something wrong when configuring that serial console. everything looks fine [10:56] anything on the monitor btw ? [10:56] 8since you dropped the lcd args) [11:01] testing now [11:04] no, still nothing there. And still no login terminal. [11:04] hmm [11:04] perhaps that serial-console script on the web page parses my first console= line, thus missing the correct options for the ttyS0 [11:05] I have two, console=tty0 console=ttyS0,etc [11:05] the first console line defines the console the kernel dumps dmesg to ... [11:05] the second one defines whats used for userspace [11:05] oh [11:05] the first one is console=tty0, so it should dump dmesg to the screen then [11:05] your second one say console=ttyS0 ? [11:06] console=ttyS0,115200n8 [11:06] * ogra_ wonders if the armada actually defines the serial console as S0 [11:06] try dopping *all* console args for a test [11:06] hm [11:06] ok [11:06] and see what happens :) [11:08] though looking at the cubox wiki they all seem to use a single console arg defined as console=ttyS0,115200n8 [11:09] yeah, that's where I got it from [11:09] but you're right, perhaps I should drop the first console at least (trying without all now) [11:10] ok, so removing all made screen stop receiving output after "Uncompressing Linux... done, booting the kernel." [11:10] I'll try just the first one, console=ttyS0,115200n8 now [11:11] right, but did you see anything on the monitor ? [11:11] no, I checked that too. tried to switch tty as well (ctrl+alt+fX) [11:11] (and make sure to wait for a while, it might take a bit to get to the login prompt, giv it time) [11:12] ... that I didn't do. will redo that attempt after trying the single console= [11:12] its an 800MHz system booting off slow media after all [11:12] true [11:13] "After a while (it takes a bit) a framebuffer console will appear, hiding the boot messages. Don't be scared, and wait for a few minutes. The login prompt will finally appear. " [11:14] from the wiki :) [11:14] this is why us old school guys like out console messages, gives us hope :-D [11:14] yeah [11:14] :) [11:15] well, I think the framebuffer console won't appear here, this is not the stock kernel. The 2.6 kernel had graphic drivers and used a second command line arg, video= with resolution and stuff [11:15] what kernel is it then ? [11:15] 3.3.1 [11:15] and are you 100% sure that kernel is supposed to work at all [11:15] (with ubuntu) [11:16] compiled by me, config from forum post where they said it booted fine [11:16] on ubuntu ? [11:16] they had Debian [11:16] perhaps it just wont work? [11:16] well, the debian kernel should tehoretically work [11:17] http://www.solid-run.com/phpbb/viewtopic.php?f=9&t=493 [11:17] this is the link where this user (which doesn't respond back) says he got it working with stock 3.3.1 kernel [11:17] and Debian 6 [11:17] so I figured Ubuntu Core would work with that kernel as well [11:18] first tried booting Ubuntu Core with kernel 2.6, but got looping error messages from dmesg complaining about missing functionality [11:19] the kernel I used is downloaded from www.kernel.org [11:19] ogra_: all three work top, premount and bottom [11:20] lilstevie, well, then blame usersapce or some kernel option the userspace needs [11:20] heh [11:20] but I'm still in no better position to debug :/ [11:20] but if the serial-console would have been working, I should have been given a login prompt in my screen session to /dev/ttyUSB0, correct? [11:20] "Also note that currently, the "make modules" fails, there is some sort of compile error for the framebuffer modules. I just ignored them at initial testing...." [11:21] so why would you expect anything on your monitor at all ? [11:21] ogra_: yeah, I've read that. I figured that was the framebuffer module (with support for higher resolution and such), and that it would fallback to just displaying the console using some basic graphics drivers [11:22] but perhaps you're right, perhaps it just doesn't work at all [11:22] but I still should get a login console on the serial-console [11:23] true [11:24] can you pastebin your full output from the screen session ? [11:26] (from one boot attempt indeed) [11:26] hmm, trying to get the whole screen buffer [11:26] hardcopy gives me the current window [11:28] I'm on it [11:32] where do screen logs go with the -L option? [11:33] C-a H (log) Begins/ends logging of the current window to the file "screenlog.n". [11:36] http://pastebin.com/NqfgpiCT [11:39] ogra_: I have to leave for a few hours, will be back here later. I am really grateful for your help, I've been stuck on this problem for some days now and it's frustrating when I don't know even where to start looking [11:39] yep [11:39] i might not be around though (i usually dont work on weekends :) ) [11:40] i wonder what /etc/init,conf is supposed to be [11:41] the /etc/init.conf in the rootfs? [11:41] they don't mention that one in the OMAP guide [11:41] well, it usually doesnt exist on ubuntu systems [11:42] but seems ot be read on yours [11:42] hmm.. something specific to ubuntu core perhaps? [11:43] maybe upstart doesn't properly start my serial-console script because of some errors with /etc/init or /etc/init.conf [11:43] as i said, /etc/init.conf shouldnt exist [11:44] and it seems the boot stops at upstart-udev-bridge [11:44] probably soemthing you should take to our upstart maintainer on monday [11:44] (he is jodh in #ubuntu-devel) [11:44] actually, when I mount the root partion on another computer and look in /etc/ the file isn't there (init.conf) [11:45] yes, as its supposed to be, not sure why there is that message in your dmesg [11:45] true, maybe to notify me that if the file had been there, it should have been read [11:46] I do have the --verbose option on [11:46] yeah, likely [11:46] in any case it doesnt seem to even get to fire up a getty process [11:47] either a kernel misconfiguration or some issue with upstart [11:47] ok [11:47] then I have a pointer on where to continue next [11:47] thanks a lot for all your help! [11:47] np :) [11:47] I'll report back here on how it goes :) [11:48] really appreciate your help, worth a lot for me. take care! [12:15] ok ogra_ I can trace it to something in bottom [12:15] I get a panic some boots with break=bottom [12:46] Are there no mirrors of Ubuntu ports? [14:42] ogra_, you here? [14:51] ogra_: I got it working!!! the reason I didn't get a console was that I had accidentally placed the serial-auto-detect-console.conf in /etc/, not in /etc/init [14:51] !!! [14:52] thanks again for all your help! now that I have a root console, I'm in my safe place again :) [18:37] ogra_: in case anyone else comes in asking questions about kernel 3.3.1 and ubuntu core :) http://www.solid-run.com/phpbb/viewtopic.php?f=4&t=544 [18:54] jeinor: Not that it matters for kernels that are ABI-agnostic, but if you're going to recommend people install an ARM cross-compiler, recommend they install gnueabihf, please. [18:54] We really want to encourage people to switch to armhf completely. [18:55] jeinor: Err, and same for core itself, you tell people to use armel, again, they'll be much happier with armhf. [19:21] Il [19:21] wops, I'll change that [19:21] infinity, what is ABI-agnostic? [19:22] to run a ubuntu core armhf, will I need to recompile kernel using the gnueabihf cross compiler? [19:24] more importantly, will it make any performance difference? [19:24] jeinor: No, the kernel doesn't care what ABI your userspace is at all. [19:25] jeinor: Using either compiler will get you an identical kernel. [19:25] jeinor: I'd just prefer we encourage people to use hf in all cases. [19:32] I'll test that for my install and if everything works as before I'll change it in the post [19:32] thanks for pointing it out [19:41] hmm, what happens to all the armel-packages out there? in apt repositories for example? can I run armel compiled binaries on armhf systems? [19:41] With multiarch, you might be able to. [19:41] But generally, you should compile things for armhf. [19:42] They're ABI-incompatible. [19:42] The question is basically the same as "can I run i386 binaries on amd64", to which the answer is "sometimes". [19:43] I see. What about the packages in apt repositories? Are most armhf or armel nowadays? [19:43] openjdk, for example? [19:47] i think, there are two ports. one for armel and one for armhf [19:52] and as far as i know, the armhf port was completely rebuild (some weeks ago?). openjdk is also in the armhf-universe-repo of precise [19:53] ok, thanks [19:55] when you're saying "the armhf port was completely rebuilt", do you mean all packages in the official ubuntu repo were rebuilt for precise? [19:55] for armhf in generell, i think [19:59] sounds promising, I'll see what packages are there when I begin to setup my system [20:01] it's quite eays to geht the package-information: download http://ports.ubuntu.com/dists/precise/universe/binary-armhf/Packages.gz for main/restricted/universe/multiverse and grep for the "Package: " [20:02] * row [20:04] nice, thanks for the tip [20:04] np, you are welcome [20:06] jesus, my grammar is bad today [20:58] jeinor: There are two different prices ports. armhf and armel. armhf is the one we're supporting. [22:04] jeinor, http://packages.ubuntu.com === Ursinha` is now known as Ursinha === Ursinha is now known as Guest65814 [22:44] scientes, there is only i386, amd64 and powerpc selectable [22:45] oh yeah ubuntu has an ugly split unlike debian [22:45] there is another site, but i forget where it is