=== lilstevie|ZNC is now known as lilstevie | ||
RobotGuy | I've configured and built a new custom kernel for Ubuntu, but am having difficulty getting it to boot. There is at least one error in the rebuild scrpt in the tools directory - it doesn't pass the correct kernel version (of the new kernel) to the commands. I'm unable to use a keyboard attacked to the Beagle-xM to get more info. | 05:03 |
---|---|---|
RobotGuy | Kernel drops me to an (initfs) prompt, but I can't enter anything to get more info. | 05:04 |
RobotGuy | This is under Maverick 10.10 | 05:08 |
=== JaMa|Off is now known as JaMa | ||
=== hrw|gone is now known as hrw | ||
hrw | hi | 08:48 |
=== JamieBen1ett is now known as JamieBennett | ||
=== ian_brasil__ is now known as ian_brasil | ||
egost | hi all | 10:47 |
egost | i have troubles getting PowerVR SGX driver to work on blaze (omap4 ES1.0). | 10:47 |
egost | can somebody confirm that? | 10:47 |
egost | PVR_K: (FAIL) SGXInit: Incompatible HW core rev (10100) and SW core rev (10200). | 10:48 |
egost | whole dmesg output : http://pastebin.com/raw.php?i=ZzhF7Lzy | 10:50 |
egost | var/log/Xorg.0.log : http://pastebin.com/raw.php?i=w58UUAkU | 10:52 |
=== ian_brasil__ is now known as ian_brasil | ||
=== arun__ is now known as arun | ||
=== kmargar is now known as markos_ | ||
bernard_ | rsalveti: so, i couldn't reproduce the sd card corruption today, which is somewhat mysterious. | 13:01 |
bernard_ | i swear it happened repeatedly on monday, but maybe i was on something good. | 13:01 |
bernard_ | on an unrelated note, could this patch be incorporated into the ubuntu omap4 kernel? | 13:02 |
bernard_ | http://marc.info/?l=linux-omap&m=128744421925804&w=2 | 13:02 |
bernard_ | it's upstreamed and in 2.6.37-rc1 | 13:02 |
* hrw wants it | 13:03 | |
ogra_ac | how would that affect systems where users set the MAC from u-boot already ? | 13:03 |
hrw | anyway my dhcp server forces same ip for device which asks as panda | 13:04 |
bernard_ | in theory it shouldn't affect that at all. | 13:04 |
bernard_ | it just allows you to override it after boot. | 13:04 |
bernard_ | does the ubuntu omap4 kernel have a different patch for setting the mac address? | 13:04 |
ogra_ac | nope | 13:04 |
ogra_ac | file a bug and ask for SRU | 13:05 |
ogra_ac | should be possible to pull it into maverick | 13:05 |
=== hrw is now known as hrw|afk | ||
bernard_ | so it seems that the "linux" source package is collecting omap kernel bugs too. is that right? | 13:07 |
ogra_ac | for omap3, yes | 13:08 |
ogra_ac | and only in maverick | 13:08 |
bernard_ | what about omap4? | 13:08 |
ogra_ac | for panda you want linux-ti-omap4 | 13:08 |
bernard_ | ah, cheers. | 13:09 |
bernard_ | mind you, the beagle has the same ethernet chip. | 13:09 |
ogra_ac | which is fine given we have two different kernel trees | 13:10 |
ogra_ac | for natty omap3 will use upstream | 13:11 |
bernard_ | so should i file a bug on the linux package too? | 13:11 |
ogra_ac | for maverick if you want to | 13:11 |
bernard_ | ah k | 13:11 |
ogra_ac | natty will get it qutomatically | 13:11 |
bernard_ | where does the SRU tag go? subject? or is there a magic setting? | 13:11 |
ogra_ac | https://wiki.ubuntu.com/StableReleaseUpdates | 13:12 |
bernard_ | wow. i couldn't even get two consecutive bug numbers within a couple of minutes. that's a disturbing bug rate :/ | 13:26 |
bernard_ | but done. thanks! | 13:26 |
ogra_ac | well, we have many users ;) | 13:28 |
bernard_ | many happy users too, i'm sure :) | 13:30 |
ogra_ac | hopefully :) | 13:30 |
* bernard_ loves having a native ARM build environment where apt-get just works. | 13:31 | |
rsalveti | bernard_: what happen when you set the mac address at the cmdline? | 13:32 |
rsalveti | does it works as expected and never reset it again? | 13:32 |
bernard_ | rsalveti: cmdline as in kernel cmdline? | 13:33 |
rsalveti | bernard_: yup, the current supported method | 13:33 |
rsalveti | the patch seems fine, but it's something that you can already work with if you set it up at the kernel cmdline | 13:35 |
bernard_ | okay, to save me the 15 minutes to download the kernel source, what's the kernel parameter? :) | 13:42 |
rsalveti | 1 sec, changing git branch :-) | 13:43 |
rsalveti | macaddr=01:23:45:67:89:AB | 13:44 |
rsalveti | for example | 13:44 |
rsalveti | so, I agree it makes sense to apply your patch, but the reason would be more to avoid recreating the mac address on every ifdown/ifup | 13:45 |
rsalveti | because we already have a way to stick with only one mac address | 13:45 |
bernard_ | cat /proc/cmdline | 13:45 |
bernard_ | vram=16M,omapfb.vram=0:5M,1:5M mem=460M@0x80000000 mem=256M@0xA0000000 root=/dev/sda1 fixrtc console=ttyO2,115200n8 rootwait macaddr=00:80:c8:40:8a:de | 13:45 |
bernard_ | but: % ip link list usb0 | 13:45 |
bernard_ | 2: usb0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1492 qdisc pfifo_fast state UNKNOWN qlen 1000 link/ether 7a:ea:e9:22:5a:4b brd ff:ff:ff:ff:ff:ff | 13:45 |
bernard_ | uname -a: Linux cachehit 2.6.35-903-omap4 #14-Ubuntu SMP PREEMPT Wed Oct 6 17:23:24 UTC 2010 armv7l GNU/Linux | 13:47 |
rsalveti | http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-maverick.git;a=commitdiff;h=10f38b455e75b85f72e98786e5518cf7b0324634;hp=f62e143182cc123fdfdf9bb88952a938af7d86e8 | 13:47 |
bernard_ | rsalveti: another reason for using that patch is that it's what's in upstream, and consistent with other drivers. it also shouldn't break people who set it with current methods. | 13:47 |
rsalveti | bernard_: sure | 13:48 |
bernard_ | ahh. i think it needs smsc95xx.macaddr then | 13:48 |
rsalveti | sure, you're right | 13:48 |
rsalveti | ok, will build a new kernel with your patch, and will put this comment at the bug | 13:49 |
rsalveti | thanks for reporting it | 13:49 |
bernard_ | np. thank you! | 13:49 |
bernard_ | ack. the smsc95xx.macaddr= indeed works too. | 13:52 |
rsalveti | cool | 13:52 |
sebjan_ | rsalveti, bernard_: just reconnecting to irc right now and see your chat:) | 14:05 |
sebjan_ | I have already commented on the launchad bug | 14:05 |
rsalveti | sebjan_: after posting my comment I saw that you also commented on it :-) | 14:05 |
bernard_ | snap! | 14:06 |
rsalveti | sebjan_: I'm now creating a new deb file to be tested | 14:06 |
bernard_ | there is also bug 673509 for omap3 | 14:07 |
ubot2 | Launchpad bug 673509 in linux (Ubuntu) "Beagleboard-xm chooses a new IP address on each boot (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/673509 | 14:07 |
rsalveti | true, this would affect both | 14:07 |
bernard_ | (sorry to make you do more work! know that it's all greatly appreciated!) | 14:08 |
rsalveti | one more to the pipeline | 14:08 |
rsalveti | :-) | 14:08 |
rsalveti | np at all | 14:08 |
=== hrw|afk is now known as hrw | ||
ogra_ac | NCommander, so i traced down the image build issues we have atm with cjwatson | 14:23 |
ogra_ac | NCommander, seems we have a dpkg bug | 14:23 |
ogra_ac | (telling you that because you wanted to work out why images fail) | 14:23 |
* rsalveti lunch | 14:40 | |
=== zyga is now known as zyga-lunch | ||
rsalveti | ogra_ac: any chance to fix the current alsa-utils issue? | 16:25 |
ogra_ac | rsalveti, not yet, i discussed several solutions and have to pick one | 16:25 |
ogra_ac | generally we need a script that removes itself after first boot and calls the alsactl init | 16:25 |
ogra_ac | s/boot/reboot/ | 16:26 |
ogra_ac | i'm just still pondering about the implementation details since all solutions are butt ugly | 16:26 |
rsalveti | ogra_ac: but we still need to explicit call alsactl init? | 16:26 |
ogra_ac | yes | 16:26 |
ogra_ac | once | 16:26 |
ogra_ac | and you need to call it after reboot | 16:26 |
ogra_ac | since the kernel we get at the same upgrade needs to be running first | 16:27 |
rsalveti | but the init is already calling when the file /var/lib/alsa/asound.state is not there | 16:27 |
rsalveti | but I didn't check if this file is created even with the wrong configuration | 16:28 |
ogra_ac | that file is always there | 16:28 |
ogra_ac | just not on brandnew installs | 16:28 |
ogra_ac | it gets created on every shutdown | 16:28 |
ogra_ac | so the code you look at is for new installs | 16:29 |
ogra_ac | it doesnt help on upgrades | 16:29 |
rsalveti | ok, it worked for me because at the first boot I just installed the new kernel and the patched alsa-utils that avoid calling alsactl init | 16:29 |
rsalveti | then at the next boot everything was working fine | 16:29 |
hrw | race condition | 16:29 |
ogra_ac | no | 16:29 |
hrw | what about systems like mine when upgrade of kernel != reboot? | 16:29 |
ogra_ac | no race condition, expected behavior | 16:29 |
ogra_ac | hrw, then a script has to be in place taking care for that | 16:30 |
hrw | I sometimes do few kernel updates before reboot | 16:30 |
ogra_ac | thats what i was talking about above | 16:30 |
ogra_ac | thats fine | 16:30 |
ogra_ac | the script will check for the right kernel running on boot | 16:30 |
ogra_ac | if it sees the right one, it will call alsactl init | 16:31 |
ogra_ac | and renove itself | 16:31 |
ogra_ac | *move | 16:31 |
rsalveti | don't know if removing itself is the best choice | 16:31 |
rsalveti | maybe creating some kind of stamp somewhere | 16:31 |
hrw | so situation is: kernel has proper alsa setup, but users have wrong one so we need to grab kernel one before alsa will use inproper one? | 16:31 |
ogra_ac | rsalveti, that would mean you check for the stamp on every boot | 16:31 |
rsalveti | ogra_ac: sure, something like what is already happening at [ -f /var/lib/alsa/asound.state ] || alsactl init | 16:32 |
ogra_ac | rsalveti, right, we shouldnt add another one | 16:32 |
rsalveti | just because I don't like stuff being removed from my system without any package removal | 16:33 |
ogra_ac | better have a link somewhere that executes it and remove that link once it was run once | 16:33 |
ogra_ac | as i said, its just about implementation details atm, i discussed all that at UDS with the audio guys | 16:33 |
rsalveti | hrw: users need to update the kernel and then call alsactl init to initialize the correct config | 16:33 |
rsalveti | so to make the sound work you need to reboot anyway | 16:34 |
ogra_ac | right | 16:34 |
ogra_ac | the kernel has the name changes to the alsa devices we require | 16:34 |
ogra_ac | without the new names the alsa-utils changes cant work | 16:34 |
ogra_ac | thats the catch22 | 16:34 |
hrw | no more "SDP4430 Media"? | 16:35 |
ogra_ac | rsalveti, persia's suggestion was a here document from postinst, so it wont be handled by any package at all, neither the putting in place nor the removal | 16:35 |
rsalveti | ogra_ac: sorry If I already asked this question, but why alsactl init is not called on every boot anymore? | 16:35 |
ogra_ac | because thats resetting the mixers | 16:35 |
hrw | rsalveti: what if you setup volume? | 16:36 |
ogra_ac | hrw, no, no more SDP4430 Media | 16:36 |
rsalveti | ok, the restore should be called, not init | 16:36 |
hrw | omap4 mixer is not small beast | 16:36 |
rsalveti | and restore also calls init when it fails | 16:37 |
ogra_ac | rsalveti, right, thats the core bug | 16:37 |
ogra_ac | restore should call init but it doesnt with our sound driver | 16:37 |
ogra_ac | that will be investigated for natty | 16:37 |
rsalveti | hm, ok | 16:37 |
ogra_ac | hrw, SDP4430 is used in more boards and all of them require different mixer setups so we added names pre board for the device | 16:38 |
ogra_ac | *per | 16:38 |
ogra_ac | that way tha alsa init scripts can adjust them on a per board base | 16:38 |
hrw | ogra_ac: I remember that discussions | 16:39 |
GrueMaster | *Hopefully* UCM will handle this in the future. | 16:39 |
rsalveti | hrw: http://paste.ubuntu.com/529441/ current result for panda | 16:39 |
ogra_ac | GrueMaster, it will, you really should have been at plumbers | 16:40 |
hrw | rsalveti: which ver of kernel? | 16:40 |
rsalveti | previously was just SDP4430 - SDP4430 for all boards | 16:40 |
GrueMaster | Yea, I figured as much. | 16:40 |
* hrw needs to revert to natty on panda | 16:40 | |
rsalveti | hrw: 2.6.35-903-omap4 #17-Ubuntu | 16:40 |
ogra_ac | revert ? | 16:40 |
rsalveti | latest one available at maverick | 16:40 |
hrw | ogra_ac: yes. today was linaro 10.11 testing day which is maverick. yesterday it was running natty | 16:41 |
ogra_ac | ah | 16:41 |
hrw | I do not have maverick on devices | 16:41 |
* ogra_ac has only maverick on all systems here | 16:42 | |
hrw | but will reinstall smartbook to maverick to check kde4/armel status | 16:42 |
* hrw reboots to #17 kernel | 16:48 | |
hrw | got panda | 16:51 |
ogra_ac | now installinf alsa-utils should get you working sound | 16:53 |
hrw | I did 'alsactl init' by hand | 16:54 |
ogra_ac | that wont give you the right setup if you dont have the proper alsa-utils | 16:54 |
hrw | ok | 16:55 |
rsalveti | ogra_ac: who calls /sbin/alsa-utils start? | 16:59 |
ogra_ac | rsalveti, udev iirc | 16:59 |
rsalveti | hm, ok | 16:59 |
ogra_ac | yeah, through an additionl script | 17:00 |
ogra_ac | from /lib/udev/rules.d/80-alsa.rules | 17:00 |
rsalveti | yup, saw it here now | 17:00 |
ogra_ac | i have a skeleton for the postinst stuff, i'll finish that tomorrow during work hours | 17:01 |
rsalveti | so it calls when it add controlC | 17:01 |
ogra_ac | right | 17:01 |
rsalveti | ogra_ac: sure, np, I'm just trying to understand now why restore is not following init | 17:01 |
ogra_ac | i think its in alsactl itself | 17:01 |
ogra_ac | look at the source there | 17:02 |
rsalveti | yeah, doing that right now | 17:02 |
ogra_ac | but generally init shouldnt be needed at all | 17:02 |
=== zyga is now known as zyga-gone | ||
ogra_ac | the driver should properly init and will hopefully do in natty | 17:02 |
rsalveti | sure | 17:03 |
ogra_ac | btw, what happened to the backtrace GrueMaster wanted to supply | 17:04 |
rsalveti | don't know | 17:04 |
ogra_ac | would be intresting to see the offending function | 17:04 |
rsalveti | but it's probably something that the compiler optimized for neon | 17:04 |
rsalveti | as neon is used for everything | 17:04 |
ogra_ac | yeah | 17:04 |
rsalveti | and not just neon capable files | 17:04 |
=== hrw is now known as hrw|gone | ||
aju | hi all,how create cross toolchain for ARM from scratch on ubuntu 10 | 17:23 |
hrw|gone | aju: "apt-get install arm-linux-gnueabi-gcc" is not enough? | 17:24 |
hrw|gone | ;d | 17:24 |
aju | hrw|gone, want to know complete process for creating toolchain on ubuntu 10 | 17:26 |
hrw|gone | aju: apt-get source armel-cross-toolchain-base + build it + install results + apt-get source gcc-4.5-armel-cross + build it + install it + apt-get source gcc-4.4-armel-cross + build it + install it | 17:27 |
* hrw|gone out | 17:27 | |
RobotGuy | How do I replace the kernel in Ubuntu? I've built a 2.6.36 kernel and I want to install it. | 17:46 |
RobotGuy | Ubuntu 10.10 built from an image. | 17:47 |
aju | dpkg -i | 17:47 |
aju | RobotGuy, using dpkg -i | 17:47 |
aju | \pkg name | 17:47 |
RobotGuy | I am talking building the kernel from sources, not installing from a binary. | 17:47 |
ogra_ac | https://wiki.ubuntu.com/KernelTeam/GitKernelBuild | 17:47 |
=== robbiew is now known as robbiew-lunch | ||
RobotGuy | I think you are not understanding my question.. I have configured and built a 2.6.36 kernel from sources. I want to install the new kernel and boot it. How do I do that? | 17:51 |
aju | RobotGuy, now you have .deb package of newly compiled binary? | 17:51 |
aju | i mean for kernel 2.6.36 ? | 17:52 |
RobotGuy | aju: NO I do not have a .deb package for the kernel. | 17:52 |
ogra_ac | RobotGuy, you should build a deb | 17:52 |
ogra_ac | if you know *ecxactly* wht your specific board needs for kernel and initrd treatment you can do that by hand indeed | 17:52 |
ogra_ac | but i would refer to the board documentation for that | 17:53 |
RobotGuy | As I said, I built a kernel from sources - 2.6.36 I want to install my new kernel and boot it. | 17:53 |
aju | RobotGuy, yes either you can create .deb using make-kpkg and then use chroot and install it in target board or using ssh | 17:54 |
RobotGuy | You are making this way over complicated. | 17:55 |
ogra_ac | RobotGuy, then look at your board docs what your board needs exactly | 17:55 |
ogra_ac | i.e. a u-boot booting board needs a uImage made from vmlinuz | 17:55 |
RobotGuy | I have already built a kernel for my hardware. Now I want to install it and boot it. What is so difficult here? | 17:56 |
ogra_ac | a redboot based board needs special treatment to dd the kernel into right place | 17:56 |
rsalveti | RobotGuy: the ubuntu way would be to have a package for it | 17:56 |
ogra_ac | that you didnt tell us anything about your hardware yet | 17:56 |
rsalveti | if you already built your own kernel you can just give make deb-pkg | 17:56 |
rsalveti | the upstream kernel also has a way to build a deb from the sources | 17:56 |
rsalveti | and it works quite well with ubuntu too | 17:56 |
RobotGuy | How is this going to help me install my new kernel? | 17:56 |
ogra_ac | what hardware do you have there ? | 17:57 |
RobotGuy | Hardware is a BeagleBoard-xM | 17:57 |
rsalveti | it's easier because if you're installing it on a running board, the initrd will be generated when installing it | 17:57 |
ogra_ac | so you need to run the right mkimage command | 17:57 |
ogra_ac | and create a uImage | 17:57 |
ogra_ac | then put it in the right place | 17:57 |
RobotGuy | I already have a uImage | 17:57 |
rsalveti | RobotGuy: if you just want to use a new uImage, just put it at the right place at the first partition | 17:57 |
rsalveti | and make sure your modules are located at /lib/modules | 17:57 |
ogra_ac | generate a corresponding uInitrd and put that in place as well | 17:57 |
RobotGuy | rsalveti: That does not work. | 17:58 |
ogra_ac | then your mkimage parameters were likely worng | 17:58 |
ogra_ac | where/how does it stop ? | 17:58 |
rsalveti | RobotGuy: it should work fine, if you put the modules in place and call depmod -a | 17:58 |
ogra_ac | a deb cares for all that | 17:59 |
RobotGuy | Is there a .deb package that matches a kernel 2.6.35.7-16, including config?? | 17:59 |
rsalveti | RobotGuy: having the deb file is easier, you just need to install it | 17:59 |
rsalveti | and then making sure you're booting with the new uImage and uInitrd | 17:59 |
RobotGuy | rsalveti: You do not understand. I need sources to match the running kernel for development purposes. | 17:59 |
ogra_ac | flash-kernel will care in this case | 18:00 |
rsalveti | RobotGuy: so just use the new uImage, copy the modules to /lib/modules, update the modules cache with depmod and generate a new initrd | 18:00 |
rsalveti | that should also work | 18:00 |
RobotGuy | rsalveti: I have already tried that and it does not work. | 18:01 |
aju | which emulator or IDE is available for ubuntu armel | 18:01 |
ogra_ac | qemu | 18:01 |
rsalveti | RobotGuy: it should work fine, probably something when wrong | 18:01 |
aju | i mean i like test my app | 18:01 |
ogra_ac | RobotGuy, well, if you dont give us any more info we cant help | 18:01 |
aju | ogra_ac, for qemu need to boot complete os ? | 18:02 |
ogra_ac | what does happen exactly or what doesnt ? | 18:02 |
aju | or just application ? | 18:02 |
ogra_ac | aju, both | 18:02 |
RobotGuy | What more info do you need? | 18:03 |
ogra_ac | you can install qemu-arm-static and run qemu-debootstrap --arch armel to cerate an arm chroot | 18:03 |
ogra_ac | or you can use qemu-system, grab a versatile kernel and run a full VM | 18:03 |
rsalveti | RobotGuy: did you give make modules_install INSTALL_MOD_PATH=/somewhere and then copied the modules to the correct directory? | 18:03 |
ogra_ac | <RobotGuy> rsalveti: That does not work. | 18:04 |
ogra_ac | <ogra_ac> then your mkimage parameters were likely worng | 18:04 |
ogra_ac | <ogra_ac> where/how does it stop ? | 18:04 |
RobotGuy | rsalveti: I know how to build kernels. I've been doing it for years. | 18:04 |
rsalveti | RobotGuy: so please post the exact error you're getting | 18:05 |
RobotGuy | rsalveti: My problem does not have to do with building the kernel. There is no error to post. | 18:05 |
rsalveti | RobotGuy: you say it didn't work for you, but didn't say what went wrong | 18:06 |
RobotGuy | First of all, I never said anything went wrong. Secondly, I came here asking how to do something, not how to correct an error. | 18:07 |
rsalveti | RobotGuy: well, I gave you two working methods, and you said you don't want to use deb and the other one simply doesn't work for you, for some unknown reason | 18:08 |
rsalveti | so I don't know how to help you more, unless you post us what went wrong with one of these options | 18:09 |
RobotGuy | rsalveti: I told everyone what I need to do at the start. Nobody has told me how to accomplish what I want to do. | 18:09 |
rsalveti | RobotGuy: I told you 2 solutions, one is to build a deb and another is to produce a new uImage, install the modules by hand, generate the initrd and then reboot | 18:10 |
rsalveti | as you asked how to replace ubuntu's kernel with your own sources | 18:10 |
RobotGuy | I asked how to replace Ubuntu's kernel with my own kernel that I built from sources. | 18:13 |
RobotGuy | This should not be rocket science. | 18:14 |
rsalveti | and I gave you two working solutions for that | 18:15 |
GrueMaster | RobotGuy: Let's start over. Are you using our pre-installed image? Is that the kernel you wish to replace? Do you want to replace it before or after first boot? | 18:15 |
RobotGuy | I started out from here ---> http://elinux.org/BeagleBoardUbuntu | 18:16 |
* GrueMaster looks. | 18:17 | |
RobotGuy | I installed the Ubuntu image from there - 10.10 | 18:18 |
GrueMaster | Ok, that lists several images, both Ubuntu supported and some that someone created from ubuntu repos and posted as tarballs. It also lists different versions (Maverick, lucid). | 18:18 |
RobotGuy | I want to replace the kernel. | 18:18 |
RobotGuy | maverick | 18:18 |
GrueMaster | Please be more specific. We have a pre-installed image that https://wiki.ubuntu.com/ARM/OMAPMaverickInstall points to, and there is a tarball on the link you posted as well. | 18:19 |
GrueMaster | Two entirely different beasts. | 18:20 |
GrueMaster | The "Demo Image" is not built or supported by Ubuntu other than at the package level. The ubuntu-netbook-10.10-preinstalled-netbook-armel+omap.img.gz that can be downloaded from http://cdimage.ubuntu.com/ubuntu-netbook/ports/releases/ is supported. | 18:22 |
RobotGuy | All I want to do is have sources matched to the kernel my beagleboard is running. I need this for development. | 18:22 |
GrueMaster | Either way, we need to know exactly which one because they are completely different and require different approaches to replace the kernel. | 18:23 |
RobotGuy | Which what?? | 18:23 |
GrueMaster | Which image. | 18:23 |
RobotGuy | I already gave the info on where I got the image. | 18:23 |
RobotGuy | http://elinux.org/BeagleBoardUbuntu | 18:24 |
RobotGuy | Maverick 10.10 | 18:24 |
GrueMaster | I am looking at that link. There is Maverick 10.10 preinstalled (linked through https://wiki.ubuntu.com/ARM/OMAPMaverickInstall), and below it is the Demo Image based on Maverick 10.10. | 18:24 |
GrueMaster | So again, which image are you using? | 18:25 |
RobotGuy | Demo Image, Maverick 10.10 | 18:25 |
GrueMaster | Ok. Thank you. Now we have a starting point. | 18:25 |
GrueMaster | Unfortunately, I have know knowledge of that image at this time, but I think rsalveti's point of building a kernel deb package and installing that is the best bet. | 18:27 |
RobotGuy | WHY do I want to build a kernel deb. That is an unnecessary complication to what I need to do. | 18:28 |
GrueMaster | If you already have a vmlinuz & initrd.img, then you can alternatively use mkimage to make a uImage and uInitrd to boot from. | 18:28 |
aju | i have a basic question why uImage , u boot image x loader need to copy in Fat partition why not in ext ? | 18:29 |
RobotGuy | I already have a uImage. I need to create the matching uInitrd. | 18:29 |
GrueMaster | If you scroll down the web link you posted, it has how to create a uInitrd from initrd.img. | 18:29 |
RobotGuy | My basic question remains though: How do I get the sources that match the kernel that is installed with the image I am using?? | 18:30 |
GrueMaster | By installing the deb packages created when building the binaries. Or copying the kernel source to /usr/src. | 18:31 |
RobotGuy | The whole reason I went to build my own kernel is that I could not get sources to match the kernel on the image. | 18:31 |
ogra_ac | RobotGuy, ask the person that created the image you used | 18:31 |
ogra_ac | you are obviously not using an ubuntu image | 18:31 |
RobotGuy | I'm using an image with ubuntu on it. | 18:32 |
ogra_ac | right | 18:32 |
ogra_ac | but not an image created by any ubuntu developer or in the ubuntu infrastructure | 18:32 |
GrueMaster | As I said, we can support it at the *package* level (i.e. if there is an ubuntu package in it that is having issues). | 18:32 |
ogra_ac | we can give you advise for ubuntu images | 18:32 |
ogra_ac | or for ubuntu packages | 18:33 |
RobotGuy | How do I know what images are or are not supported? | 18:33 |
GrueMaster | If the images come from *.ubuntu.com. | 18:33 |
ogra_ac | RobotGuy, GrueMaster just pointed you to some urls | 18:34 |
GrueMaster | (or authorized mirrors). | 18:34 |
GrueMaster | Our officially supported omap image for maverick is http://cdimage.ubuntu.com/ubuntu-netbook/ports/releases/maverick/release/ubuntu-netbook-10.10-preinstalled-netbook-armel+omap.img.gz | 18:35 |
RobotGuy | Will that work on a BeagleBoard? | 18:36 |
GrueMaster | Installation instructions are at https://wiki.ubuntu.com/ARM/OMAPMaverickInstall | 18:36 |
GrueMaster | Yes. | 18:36 |
GrueMaster | Tested on Beagle C4 & Beagle XM. | 18:36 |
RobotGuy | OK, thanks. I guess I need to switch images then. I have a C3 and an xM. | 18:36 |
GrueMaster | It should also work fine on the C3. On the XM, be aware that there is a workaround for DVI on XM rev A3 or later. | 18:37 |
RobotGuy | My xM is an A2 | 18:37 |
GrueMaster | Then it should just work. | 18:37 |
aju | ogra_ac, i have a basic question why uImage , u boot image x loader need to copy in Fat partition why not in ext ? | 18:38 |
GrueMaster | A3 moved the DVI enable line to a gpio after 10.10 release. It has been fixed in a kernel update, but is tricky on first boot. | 18:38 |
GrueMaster | But as you have A2, no worries. | 18:38 |
RobotGuy | I'm glad. :D All this time I didn't know I was using an unsupported image of Ubuntu. | 18:38 |
ogra_ac | aju, thats a u-boot limitation we had with the omap u-boot | 18:38 |
GrueMaster | RobotGuy: No problem. There are a lot of images and even full distributions based on Ubuntu (i.e. Mint) packages. But we can only support the ones *we* create. | 18:41 |
ogra_ac | and indeed, instead of switching images you can just talk to the creator of the image you use | 18:42 |
ogra_ac | i.e. for a bug in mint you would talk to the mint devs | 18:42 |
aju | ogra_ac, can uboot read kernel from ext3 ? | 18:44 |
ogra_ac | ext2 | 18:44 |
aju | ok ext2 ? | 18:44 |
ogra_ac | but not all revisions | 18:44 |
RobotGuy | I don't know who created it apparently. I know you guys didn't make it now. | 18:45 |
aju | because as convention it is required to copy uboot xloader and kernel in fat | 18:45 |
RobotGuy | I'll just switch images. That seems to be the easist path. :) | 18:45 |
aju | will it work just copy uboot in fat and other in ext2? | 18:45 |
aju | something i want to test and run differently | 18:45 |
ogra_ac | you will have to rebuild u-boot with ext2 support | 18:46 |
ogra_ac | it rpovided to be unstable when we tested it | 18:46 |
RobotGuy | GrueMaster: I understand the whole support thing. ;D | 18:53 |
RobotGuy | I just created an sd card with that image but it says it doesn't have a valid partition table. | 19:00 |
RobotGuy | I forgot to gunzip it | 19:01 |
ogra_ac | you shouldnt gunzip | 19:01 |
ogra_ac | https://wiki.ubuntu.com/ARM/OMAPMaverickInstall has the proper instructions, use zcat | 19:02 |
RobotGuy | That ubuntu supported image just freezed my beagleboard up on kernel boot. | 19:25 |
GrueMaster | Do you see anything on the monitor? | 19:27 |
RobotGuy | Yeah. It appears there is a wait, reboot, wait cycle for this image to boot. | 19:56 |
RobotGuy | Ahhh, x-image and I don't have keyboard/mouse connected yet. | 19:58 |
GrueMaster | The first boot resizes the rootfs to fill the SD card. The second one should boot into X with oem-config running so that you can setup keyboard locale, timezone, and user login info. | 19:59 |
RobotGuy | It does. My misunderstanding. | 19:59 |
RobotGuy | I don't need X for the project I am working on though. | 19:59 |
rsalveti | RobotGuy: you can boot with "text" at the cmd argument and you'll not get X | 20:01 |
RobotGuy | I would prefer to not have X on the SDcard at all. | 20:02 |
RobotGuy | I need dev stuff like gcc, etc. | 20:03 |
RobotGuy | dev libraries. | 20:03 |
RobotGuy | A console only dev image would be what I need/ | 20:04 |
GrueMaster | RobotGuy: The other thing you could do is build your own image with rootstock. | 20:04 |
RobotGuy | I may have to do that. | 20:05 |
rsalveti | yep, unfortunately we don't have a minimal image for maverick | 20:05 |
rsalveti | rootstock is the way to go | 20:05 |
rsalveti | at least for natty we'll have it | 20:05 |
RobotGuy | That's too bad. | 20:05 |
RobotGuy | We lose the serial console. | 20:07 |
RobotGuy | At least if there were a minimal image, we could build up exactly what we need. | 20:08 |
GrueMaster | One other (untested) solution is to download the netboot image and use it to install from the pool. | 20:09 |
rcn-ee_at_work | RobotGuy, once you have my image it's easy to transition to ubuntu's kernel. ;) | 20:09 |
RobotGuy | rcn-ee_at_work: What image is that? | 20:10 |
rcn-ee_at_work | i was catching up on the irc log, 'my' image. ;) | 20:10 |
RobotGuy | Would that be the one I am using now? | 20:11 |
rcn-ee_at_work | Yeap it is.. from reading the backlog, you build a uImage and modules right? just dump the uImage in the fatfs and untar the modules to the rootfs.. the boot.. | 20:11 |
rcn-ee_at_work | after you build, there's a script under /boot/uboot/tools to rebuild the uInird, then reboot with all new.. | 20:12 |
RobotGuy | It can't find the linux partition | 20:12 |
RobotGuy | Yeah, I know the script. It doesn't do the right thing - uses the wrong kernel version. | 20:12 |
rcn-ee_at_work | it uses the 'kernel' version from "uname -m or -r" | 20:13 |
rcn-ee_at_work | so you need to boot with uImage for it to work.. | 20:13 |
RobotGuy | It always uses the version of the running kernel, which is wrong if I just created a new kernel. It needs to use the version of the new kernel. | 20:13 |
RobotGuy | I know that | 20:13 |
rcn-ee_at_work | yeah, so copy the 'new' uImage to the boot, then reboot with the new uImage... | 20:13 |
RobotGuy | You are not listening to me. | 20:14 |
rcn-ee_at_work | as long as your modules are populate to /lib/modules/new_image/ | 20:14 |
RobotGuy | I know how to deal with kernels, but what you are saying does not work. | 20:14 |
rcn-ee_at_work | Yeah i am.. So your running an old kernel.. You build a new one.. Take that new uImage and copy it to the fatfs dir.. then reboot your beagle, it'll run the new uImage.. | 20:14 |
RobotGuy | It does not find the rootfs partition. | 20:15 |
RobotGuy | I don't know why. | 20:15 |
rcn-ee_at_work | sounds like a config problem... can you pastebin your .config | 20:15 |
rsalveti | ogra_ac: back to the alsa-utils issue | 20:24 |
rsalveti | alsa-utils is calling alsactl restore with -I | 20:24 |
rsalveti | -I,--no-init-fallback | 20:25 |
rsalveti | it calls with no init fallback, gets the error, tries to sanify it and then calls restore again | 20:25 |
RobotGuy | rcn-ee_at_work: http://pastebin.com/87rjwxWm | 20:26 |
rcn-ee_at_work | Thanks RobotGuy | 20:26 |
RobotGuy | If it's the config, I don't know what the problem is. | 20:27 |
rcn-ee_at_work | most of the obvious things are right, but "CONFIG_GPIO_TWL4030=m" should be builtin, i belive it sets the power to mmc card... | 20:31 |
rcn-ee_at_work | diff'd mine and ubuntu's config we both got it as "=y" | 20:31 |
RobotGuy | OK. Would that cause it to not boot and not find things? | 20:32 |
RobotGuy | Yes, I can see where that would make the mmc not available | 20:32 |
rcn-ee_at_work | Well alot of the GPIO's on the TWL4030 control essential things like power control for the ehci/musb/mmc so if they don't get power... | 20:32 |
RobotGuy | Would you fix and repost it for me please? I just want it to work. | 20:33 |
RobotGuy | I hate dealing with kernel configuration | 20:34 |
rcn-ee_at_work | RobotGuy, i pushed it here: http://pastebin.com/uPS0w5dh but if you have any more issues, my config for refernence is here: http://bazaar.launchpad.net/~beagleboard-kernel/%2Bjunk/2.6-stable/annotate/head:/patches/lucid-defconfig | 20:35 |
rsalveti | ogra_ac: just tested here, when booting with the old state file, at least for panda it'll be ok because the card name is different anyway | 20:36 |
RobotGuy | Thankyou | 20:36 |
rsalveti | and will store the correct values after first reboot | 20:36 |
rsalveti | the problem will be just blaze, as the card name is the same as before the kernel patches | 20:36 |
rsalveti | and restore with -F will try to restore most it can | 20:37 |
rsalveti | ogra_ac: for panda the state file will be the same if you call alsa-utils start (with the initial wrong description for sdp4430) or call alsactl init | 20:41 |
rsalveti | when giving restore the result is the same here, so don't know why you need to call alsactl init for panda at least one time | 20:42 |
* rsalveti out for now, be back later | 20:43 | |
=== zyga-gone is now known as zyga | ||
RobotGuy | rcn-ee_at_work: Now I get an internal compiler error. :( | 21:06 |
rcn-ee_at_work | RobotGuy, ouch.. just by changing that one little config? what file did it ice on? | 21:07 |
rcn-ee_at_work | btw, which compiler? (the cross gcc from maverick's repo so far seems pretty sane..) | 21:07 |
RobotGuy | mm/dmapool.c:351: internal compiler error: in try_ready, at haifa-sched.c:3222 | 21:21 |
RobotGuy | gcc version 4.4.5 (Ubuntu/Linaro 4.4.4-14ubuntu5) | 21:22 |
RobotGuy | I am not cross building. | 21:23 |
rcn-ee_at_work | RobotGuy, weird, on my last build with that version, it didn't have any issues with mm/dmapool.c (2.6.35.7)... http://rcn-ee.homeip.net:81/dl/farm/log/COMPLETE-2.6.35.8-l7_1.0-maverick.txt | 21:25 |
RobotGuy | I'm building 2.6.36 | 21:27 |
rcn-ee_at_work | i know.. file hasn't really changed; http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.36.y.git;a=history;f=mm/dmapool.c;h=3df063706f53e6d579ec0acabace3503a471ba29;hb=HEAD | 21:27 |
RobotGuy | All I want is the source to match the kernel I am running. | 21:28 |
rcn-ee_at_work | what kernel are you running? | 21:28 |
rcn-ee_at_work | (uname -a) | 21:29 |
RobotGuy | 2.6.35.7-l6 | 21:29 |
RobotGuy | Linux tester 2.6.35.7-l6 #1 PREEMPT Wed Oct 20 01:32:27 UTC 2010 armv7l GNU/Linux | 21:29 |
rcn-ee_at_work | 2.6.35 + 2.6.35.7 from kernel.org + http://rcn-ee.net/deb/maverick/v2.6.35.7-l6/patch-2.6.35.7-l6.diff.gz + http://rcn-ee.net/deb/maverick/v2.6.35.7-l6/defconfig | 21:30 |
RobotGuy | OK, thanks. :) | 21:32 |
NCommander | ogra: what's the dpkg bug number? | 22:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!