**** BEGIN LOGGING AT Tue Aug 14 02:59:58 2012 Aug 14 03:22:55 hi, can anyone tell me is there a way to get a beaglebone running ubuntu to have its usb port appear as a usb kbd? or any links that may help would be apreciated.. Aug 14 03:23:34 i want to run some code on the beaglebone and have the output go to another computer that is blissfullly unaware its talking to another computer.. Aug 14 03:24:02 would like to output my data thru the usb keyboard interface to another PC.. Aug 14 07:25:51 ogra_: swarren pointed my to the first and second patch on https://git.reviewboard.kde.org/users/griffais which seem to fix a problem with the nvidia drivers Aug 14 07:25:58 (but for kwin-gles) Aug 14 07:40:04 marvin24, thanks, if they go upstream i guess the kubuntu team will want it ... Aug 14 07:40:56 ogra_: I thought more about the unity render problem Aug 14 07:41:12 it could be related Aug 14 07:41:30 yup Aug 14 07:41:41 but I failed to find the gles support code in unity ... Aug 14 07:42:12 is there a special branch ? Aug 14 07:42:23 lp:unity doesn't seem to contain it Aug 14 07:43:20 unity doesnt have much gles code there should be a few ifdefs though Aug 14 07:43:33 unity is just a bunch of compiz plugins ;) Aug 14 07:43:53 ah, so I need to check in the compiz code ... Aug 14 07:43:59 the heavyweight code part for gles is in compiz ... Aug 14 07:44:03 right Aug 14 07:44:25 and there gles is a gigantic ugly patch in debian/patches/ Aug 14 07:44:53 its just being merged upstream with hope that it will make feature freeze Aug 14 07:49:14 anyone know how to fix this? http://pastebin.com/CP1iAm29 MODULES=most (or anything) doesn't change the error Aug 14 07:55:23 ogra_: patch 2 seems to be included already, but we may need patch 1 Aug 14 07:55:52 ah, great ... Aug 14 07:56:36 but I'm not an expert Aug 14 07:57:02 the experts are in #ubuntu-desktop ... i just pinged smspillaz there Aug 14 07:57:05 the driver exposes EGL_KHR_image, EGL_KHR_image_base, and EGL_KHR_image_pixmap Aug 14 07:57:25 so there shouldn't be a problem at all Aug 14 07:58:52 ugh, I tried to reinstall initramfs-tools, and now of course it's stuck broken because of that error Aug 14 07:59:40 that doesnt look like an initramfs tools issue at all ... did you try running it with -v ? Aug 14 08:00:23 ogra_: http://pastebin.com/E6G0vA4W Aug 14 08:00:38 how can it not be an initramfs-tools issue, initramfs-tools is refusing to do it's thing Aug 14 08:01:02 because something is missing from your system that should be there Aug 14 08:01:36 like what? Aug 14 08:01:38 you apparently have a devicemapper handled device there ... seems devicemapper doesnt properly create a sysfs entry for it Aug 14 08:01:57 thats either a kernel or a devicemapper bug Aug 14 08:02:02 # ls /sys/block/dm-0 Aug 14 08:02:02 alignment_offset capability discard_alignment ext_range inflight queue removable size stat trace Aug 14 08:02:05 bdi dev dm holders power range ro slaves subsystem uevent Aug 14 08:02:59 could you give me some information on exactly what it's complaining about? Aug 14 08:03:25 hmm, it seems it tries to find an SD card for that entry Aug 14 08:04:10 did you actually build a raid based on mmc devices ? Aug 14 08:04:20 not raid, lvm Aug 14 08:04:25 aha Aug 14 08:04:50 well, to me it looks like you have a volume group made with an mmc that is not actually attached to the dm device Aug 14 08:04:52 ogra_: and lvm can do raid levels.... Aug 14 08:05:11 so there is no mmcblk entry in sysfs Aug 14 08:05:30 # find /sys/block/dm-0/ -name \*mmcblk\* Aug 14 08:05:30 /sys/block/dm-0/slaves/mmcblk0p2 Aug 14 08:06:58 Kevin`, well, start with filing a bug and adding all that you pasted above to it ... file it against initramfs-tools for a start, we can re-assign it later Aug 14 08:07:45 is there a way I can make it work now? this is holding me up, and now apt is in a semibroken state because of trying to reinstall it Aug 14 08:07:54 i guess lvm with SD cards isnt a very common case :) Aug 14 08:08:41 Kevin`, is your lvm actually your rootfs ? Aug 14 08:08:46 yes Aug 14 08:09:25 well, then there is not much help unless the bug is fixed ... you willk need a wporking initrd that assembles the lvm on boot, weems you dont have one at all yet Aug 14 08:09:33 *seems Aug 14 08:09:44 the system boots now Aug 14 08:10:00 the lvm used to be on usb, and initramfs-tools didn't glitch out for that Aug 14 08:10:13 oh, well, then just add an exit 0 to update initramfs and run dpkg --configure -a Aug 14 08:10:39 (dont forget to remove it later indeed) Aug 14 08:11:03 Generating Initramfs u-boot image... mkimage: Can't read /boot/initrd.img-3.2.0-1417-omap4: Invalid argument Aug 14 08:11:07 run-parts: /etc/initramfs/post-update.d//zz-flash-kernel exited with return code 1 Aug 14 08:11:10 heh Aug 14 08:11:22 apparenently the script is supposed to do something rather than exiting Aug 14 08:11:28 export FLASH_KERNEL_SKIP=1 Aug 14 08:12:04 sorry, forgot about flash-kernel :) Aug 14 08:12:20 ok, at least apt isn't confused now Aug 14 08:12:26 I still can't make an initrd for my new kernel Aug 14 08:12:36 no, not until the bug is fixed Aug 14 08:12:47 the above only worked around the apt issue Aug 14 08:15:35 btw, using an SD for an lvm will massively degrade the lvm's speed Aug 14 08:16:07 (at least if there are also normal disks involved it wioll slow them down) Aug 14 08:16:07 it's an arm device, and it's one of the newer ones that loves to use sd in place of nand Aug 14 08:17:36 no, its a manufacturer who likes to sell the dveice cheap so did pick the cheaper storage option for the board Aug 14 08:18:07 nand is usually lots and lots faster than any MMC card Aug 14 08:18:11 well sure, but everyone is doing it these days unfortunately Aug 14 08:18:27 but about ten times as expensive compared to just adding an SD slot Aug 14 08:18:32 even production devices like phones, tablets, etc. almost all dev boards. :( Aug 14 08:19:05 well, you can use very fast eMMCs if you actually build a phone ... but thats not comparable to SD cards Aug 14 08:19:23 yes it is. it's the same thing with different (or not) tweaks Aug 14 08:20:07 still pretty horrible for random io because of the block translation done outside the os Aug 14 08:20:29 still no reporting of failure status Aug 14 08:20:32 which is why we default to USB installs from quantal on :) Aug 14 08:21:11 the SD card is only used as boot device nowadays Aug 14 08:21:19 a usb spinny drive is a pretty big thing to have hanging off the device all the time. although of course I started with that before I got a fast sd card Aug 14 08:21:19 and really shouldnt be used for anything else Aug 14 08:21:57 a good SD card will max out at 16MB/s ... a USB disk will get you between 20-24MB/s Aug 14 08:22:25 bah no. a good sd card will max out at 1.5mbyte/sec (for 4k random writes) Aug 14 08:23:05 who cares about writes :) Aug 14 08:23:15 the os when doing anything? :) Aug 14 08:23:27 they are cached anyway ... reading is the issue Aug 14 08:24:16 writes still happen, and on an average sd card that does tham at 0.001mbytes/sec (not kidding.) it really affects usability Aug 14 08:24:39 https://bugs.launchpad.net/initramfs-tools/+bug/1036546 Aug 14 08:24:41 Ubuntu bug 1036546 in initramfs-tools "initramfs-tools fails for lvm on mmc" [Undecided,New] Aug 14 08:24:45 thx Aug 14 08:25:22 ogra_: so I should test this? =) Aug 14 08:25:34 xnox, if you like to ? Aug 14 08:25:43 ironically flash controllers with faster bulk read/write are normally the ones with horrible random performance Aug 14 08:25:55 i really think its quite a corner case to run lvm's off SDs Aug 14 08:26:09 (and ubiquity wont allow that ootb on qunatal installs) Aug 14 08:26:15 well... i have a panda board =) and I can re-install it with rootfs on lvm on sd Aug 14 08:26:24 but it should indeed get fixed Aug 14 08:26:52 ogra_: well... with a command line arg it does allow lvm instalations Aug 14 08:26:55 xnox, heh, have fun tricking ubiquity then, that wont be easy :) Aug 14 08:27:21 (partman will refuse to touch the partition table of the media you install from) Aug 14 08:27:28 (....which is your SD) Aug 14 08:28:04 ogra_: actually, it won't. although I was using the minimal installer, I think it only uses kernel+initrd to boot from Aug 14 08:28:05 hmm.... so --enable-partauto-lvm will do nothing =( sad Aug 14 08:28:06 you will need to create all partitioning you need in advance (without touching the first two partitions) Aug 14 08:28:37 on that note, sd hotplug doesn't work on this board. was kind of annoying when booted from usb Aug 14 08:28:38 oh, yeah, you can do it from a netboot install indeed Aug 14 08:28:54 hmmm... ok Aug 14 08:29:00 how does it not work ? Aug 14 08:29:16 if you plug in a card from i.e. your camera it doesnt get automounted ? Aug 14 08:29:25 correct Aug 14 08:29:33 well not just the automount Aug 14 08:29:38 it doesn't detect card removal/insertion Aug 14 08:29:49 it does here, stgrange Aug 14 08:29:55 * ogra_ just tried Aug 14 08:29:58 kernel version? Aug 14 08:30:12 qunatal current ... some 3.2 kernel iirc Aug 14 08:31:25 i'm using precise, might have different stuff there Aug 14 08:31:37 err, sorry, 3.4.0 actually Aug 14 08:32:01 welll, another bug to file then, linux-ti-omap4 is the package Aug 14 08:32:26 but that one requires me booting from something else to test again =p Aug 14 10:00:34 flash-kernel shouldn't attempt to identify compatible architectures by random stuff at the end of the kernel version string (say, git version info, or "+") Aug 14 10:14:06 Kevin`, well, it is designed for supported kernels ... you could file a wishlist bug in debian about it (though the 2.x version that was shipped in precise is dead now) Aug 14 11:06:38 when used only together with MLO and uboot.bin, shouldn't http://pastebin.pandaboard.org/index.php/view/49834947 run loaduimage and therefore boot the uImage kernel from the sd card on a pandaboard? Aug 14 11:22:15 LetoThe2nd, not sure what linaro sets as the default there, ask them :) Aug 14 11:23:01 ogra_: hm will do Aug 14 11:23:21 we never use images without boot.scr Aug 14 11:23:33 and there we dont use loaduimage Aug 14 11:24:16 ogra_: so you'd suggest to fixup boot.scr? Aug 14 11:24:40 well, rather /boot/boot.script, but yeah Aug 14 11:24:55 meh. Aug 14 11:25:10 or on quantal /usr/share/flash-kernel/bootscript/bootscript.omap Aug 14 11:25:41 ogra_: some documentation link on boot.scr{ipt} tinkering? Aug 14 11:29:47 apt-get source flash-kernel :) Aug 14 11:29:59 no specific documentation Aug 14 11:30:14 Kevin`_: what image did you use when you got lvm failure? Aug 14 11:30:31 we have a jenkin lvm tests failing on amd64 for servers right now Aug 14 11:31:43 xnox: the failure isn't during install, it's after Aug 14 11:31:56 also, I have a fun situation now, with update-initramfs again Aug 14 11:31:57 yeah, it installs and fails to reboot. Aug 14 11:32:06 in the jenkins test. Aug 14 11:32:08 I removed the kernel with + in the name from boot and /lib/modules etc Aug 14 11:32:17 but dpkg still thinks it has to rebuild it Aug 14 11:32:23 so it tries and fails, because it doesn't exist Aug 14 11:32:58 xnox: it boots just fine, the problem is just in initramfs-tools. although i'm not sure what would happen if you tried to initially install to it Aug 14 11:33:20 where's the list of kernels the system will try to rebuild the initrd for? Aug 14 11:33:43 ok. Aug 14 11:34:33 http://pastebin.com/yzymVf7u - I have no luck at all Aug 14 11:37:42 answer: something in /var/lib/initramfs-tools. luckily it's a file so it's indexed as a file Aug 14 11:37:54 doing a full text search of the filesystem wouldn't have been fun :) Aug 14 11:40:42 Kevin`_, how did you remove that kernel ? Aug 14 11:41:05 (did you remove the exit 0 from update-initramfs before doing it ?) Aug 14 11:41:08 I installed it with "make install", so I removed it with "rm" :) Aug 14 11:41:13 yes Aug 14 11:41:14 aha Aug 14 11:41:16 well Aug 14 11:41:32 then you need to manually remove the state file that was put in place for initramfs-tools Aug 14 11:41:46 (that would have been dont by the package scripts usually) Aug 14 11:41:50 look in /var/lib/initramfs-tools/ Aug 14 11:42:14 s/dont/done/ Aug 14 11:44:43 in case you were wondering, my hack to make it work temporarily while waiting for someone to look at the bug: http://pastebin.com/8rnK9Baj Aug 14 15:52:40 fyi, we are taking feedback on a possible pandaboard-nc variant over on the mailing list - http://groups.google.com/group/pandaboard/browse_thread/thread/74b5c6cc761d2e3c# **** ENDING LOGGING AT Wed Aug 15 02:59:59 2012