**** BEGIN LOGGING AT Fri Apr 09 02:59:56 2010 Apr 09 07:23:48 hi guys, I just noticed that lots of device nodes in /dev dir has "+" attribute. Could anybody point me to any doc/wiki about ACL implementation in Ubuntu? E.g. which CONFIG_XXX_ACL should be enabled in kernel and which user space utilities are using ACL to fine control FS or other stuffs. Thanks! Apr 09 07:24:57 * eggonlea is reading https://help.ubuntu.com/community/FilePermissions and https://wiki.ubuntu.com/ACL-OnByDefault. Apr 09 07:25:53 But these two doc seem not provide a whole view of ACL in Ubuntu. Apr 09 07:37:34 eggonlea: + attribute? Apr 09 07:37:47 how do you see that? Apr 09 07:38:52 lool, just with "ls -l" Apr 09 07:39:22 e.g. Apr 09 07:39:23 lli5@sh-dt-4513:/vobs/ubuntu$ ls -al /dev/snd Apr 09 07:39:24 total 0 Apr 09 07:39:24 drwxr-xr-x 3 root root 240 2010-03-24 11:18 . Apr 09 07:39:24 drwxr-xr-x 15 root root 4100 2010-04-02 18:48 .. Apr 09 07:39:24 drwxr-xr-x 2 root root 60 2010-03-24 11:18 by-path Apr 09 07:39:24 crw-rw----+ 1 root audio 116, 10 2010-03-24 11:18 controlC0 Apr 09 07:39:26 crw-rw----+ 1 root audio 116, 9 2010-03-30 18:12 pcmC0D0c Apr 09 07:39:29 crw-rw----+ 1 root audio 116, 8 2010-04-09 15:43 pcmC0D0p Apr 09 07:39:30 crw-rw----+ 1 root audio 116, 7 2010-03-24 11:18 pcmC0D1c Apr 09 07:39:33 crw-rw----+ 1 root audio 116, 6 2010-03-24 11:18 pcmC0D2c Apr 09 07:39:34 crw-rw----+ 1 root audio 116, 5 2010-03-24 11:18 pcmC0D3c Apr 09 07:39:36 crw-rw----+ 1 root audio 116, 4 2010-03-24 11:18 pcmC0D4p Apr 09 07:39:38 crw-rw----+ 1 root audio 116, 3 2010-03-24 11:18 seq Apr 09 07:39:41 crw-rw----+ 1 root audio 116, 2 2010-03-24 11:18 timer Apr 09 07:43:23 eggonlea: Hmm indeed Apr 09 07:44:13 This troubled us because we didn't enable ACL in kernel by default. :) Apr 09 07:44:25 Any doc to describe this? Apr 09 07:45:17 I'm wondering if we have to enabled all ACL configuration in kernel. Apr 09 07:46:38 I wonder whether it's an ACL Apr 09 07:49:51 Er, so, what's this? Apr 09 07:51:16 stracing ls and getfacl, it appears to query ACL indeed Apr 09 07:51:23 getxattr("/dev/video0", "system.posix_acl_access", "\x02\x00\x00\x00\x01\x00\x06\x00\xff\xff\xff\xff\x02\x00\x06\x00\xe8\x03\x00\x00\x04\x00\x06\x00\xff\xff\xff\xff\x10\x00\x06\x00\xff\xff\xff\xff \x00\x00\x00\xff\xff\xff\xff", 132) = 44 Apr 09 07:51:35 the + usually notates an ACL Apr 09 07:51:54 Yes; I checked the info page from ls and it says + is for ACL other than SELINUX Apr 09 07:51:55 and we definately support ACL in ubuntu Apr 09 07:52:05 not sure why its set on audio devices though Apr 09 07:52:17 When I actually query acls, I do see that there's an ACL in place, but I wonder where it comes from Apr 09 07:52:37 i was under the impression while we support them, we dont set up any ACL's by defualt Apr 09 07:52:54 e.g. getfacl /dev/video0 gives http://paste.ubuntu.com/411491/ Apr 09 07:52:57 well, i would start looking at udev rules Apr 09 07:53:20 So we need to know who's using ACL and what's the requirement to kernel. Apr 09 07:53:26 It's clear that plain owner + group differ from the ACL, but what I don't understand is why an ACL is needed at all Apr 09 07:53:43 eggonlea: So the device noces are created by udev Apr 09 07:53:54 lool, it should be. Apr 09 07:54:22 eggonlea: /lib/udev/rules.d/70-acl.rules Apr 09 07:54:46 So it seems that's for consolekit integration, misc devices get an ACL on top Apr 09 07:55:10 It seems that this is to allow access to the devices depending on who's logged in Apr 09 07:56:10 eggonlea, for the kernel just enable CONFIG_GENERIC_ACL and the _POSIX_ACL functions for each filesystem Apr 09 07:56:24 lool: Wow, it seems so. Apr 09 07:56:46 it kind of makes sense, but it's a bit scary at the same time Apr 09 07:56:54 ogra: I did not see any "acl" keyword in /etc/fstab Apr 09 07:57:05 do you think FS_ACL is needed? Apr 09 07:57:22 well, its enabled in all ubuntu kernels Apr 09 07:57:43 so to prevent later issues i would just try to build my kernel config as close to ubuntu as possible Apr 09 07:58:19 as i said above, i dont think we use ACL by default, the devices might be an exception because of console/policykit Apr 09 07:58:19 You are right. I'll check if there's any drawback or impact. Apr 09 07:59:40 eggonlea: In practice, things might work via groups, but it's best not to rely on that Apr 09 07:59:51 My user isn't in the audio group for instance Apr 09 08:00:25 Yes, that's why we have to add ubuntu user to audio group with non-ubuntu kernel. Apr 09 08:00:46 We'll enable it if nothing get impacted. Thanks! Apr 09 08:01:50 hmmm, I hope the omap kernel will have all the usb-gadget device-type drivers enabled. Apr 09 08:01:53 i think though that the audio group will persist for some releases still ... people still tend to remove pules which then forces you to use the group Apr 09 08:02:03 *pulse Apr 09 08:02:25 MUSB_HDRC (or whatever it is) needs to =y, but all the g_ether and g_audio shouldn't need that. Apr 09 08:13:10 DanaG: do you have a point to a known working USB (peripherals + OTG) on Beagleboard? Apr 09 08:13:31 A point? Not sure I get the question. Apr 09 08:13:41 I'm using rcn-ee's images, for now, anyway. Apr 09 08:13:55 I'm using ehci as host and otg as peripheral. Apr 09 08:13:59 ogra: does the config on bug 541030 work for you? It doesn't give me keyboard/mouse yet Apr 09 08:14:02 Launchpad bug 541030 in linux-ti-omap (Ubuntu Lucid) (and 1 other project) "omap kernel musb/ehci ports not enabled on beagleboard (affects: 1) (heat: 14)" [High,Triaged] https://launchpad.net/bugs/541030 Apr 09 08:14:16 DanaG: sorry, pointer Apr 09 08:14:32 I was looking for known working configs on 2.6.33 Apr 09 08:15:00 amitk, i only tried it on sunday, i think you updated the patch inbetween, the sunday build didnt work at all for me (USB wise) Apr 09 08:15:22 I've got my USB hub on ehci, and that isn't working Apr 09 08:16:11 I have weird behavior with my asix adapter: doesn't work if plugged in during boot -- has to be unplugged and replugged. Apr 09 08:16:42 amitk, i tried both, HUB and direct connection Apr 09 08:18:34 ogra: direct connection won't work, the ehci port only support high speed usb, not full speed (keyboard/mice) Apr 09 08:18:45 http://elinux.org/BeagleBoard#USB Apr 09 08:19:41 oh, thanks Apr 09 08:19:44 i didnt know that Apr 09 08:23:51 OTG port supports both, but I don't have a mini-mini (whatever it is called) USB cable Apr 09 08:24:43 i have a mini to normal adapter here Apr 09 08:26:48 hmm, i wonder how risky it is to bump compcache to 50% for all armel arches Apr 09 08:27:01 we have no subarch concept at packaging level Apr 09 08:27:17 http://trisoft.de/pics/ZHost.JPG <-- this is what I need Apr 09 08:27:21 so if i do it in the casper package it would happen for all Apr 09 08:27:44 amitk, isnt that the one with two shortened pins ? Apr 09 08:28:00 my adapter looks the same but doesnt have pin shortening Apr 09 08:28:28 or is that only necessary for nokia ? Apr 09 08:28:57 (that adapter works with the revB board i have) Apr 09 08:29:03 ogra: yeah, there are two kinds. Apr 09 08:29:08 ah, good Apr 09 08:29:19 you need the pins shortened Apr 09 08:29:23 hrm Apr 09 08:29:28 not so good thenm Apr 09 08:29:29 you could do it on the board as described in the wiki Apr 09 08:29:51 just short the J6 pad Apr 09 08:29:58 i definately dont want to solder on HW i'm relying on for getting the image done :) Apr 09 08:31:50 ogra: you will have to to get working keyboard/mouse through the OTG board, AFAIK Apr 09 08:31:56 DanaG: ^^ Apr 09 08:32:08 why ? i have a powered hub i can attach to the normal USB port Apr 09 08:32:35 and also several unpowered ones Apr 09 08:32:36 ogra: ok, try out the kernel on people and see if that works for you Apr 09 08:32:51 oh, "shortened"... I thought you meant "less long" Apr 09 08:32:52 ogra: this enabled rcn-ee's 'good' config Apr 09 08:32:53 is USB compiled in so i can test it with the server image ? Apr 09 08:32:57 "shortED". Apr 09 08:33:30 ogra: yes Apr 09 08:33:35 yeah, seems so ... Apr 09 08:33:42 * ogra just had to dig out the patch :) Apr 09 08:39:56 amitk, no go with unpowered hub Apr 09 08:40:04 * ogra digs for the powered one Apr 09 08:42:57 amitk, nor with powered ... tried both ports (OTG and normal) Apr 09 08:45:12 ogra: and did you verify that the config is identical to what rcn-ee recommended in the bug? Apr 09 08:45:32 amitk, i just took your kernel and replaced the uImage on the SD Apr 09 08:45:42 if its compiled in it should just work Apr 09 08:46:12 * ogra switches back to serial console to check dmesg Apr 09 08:46:14 ogra: please confirm the config. So I'm not crazy for not uploading the 'simple' fix Apr 09 08:46:51 I wonder if any other long-time beagleboard users have any inputs here... Apr 09 08:47:04 lrg: ^ Apr 09 08:48:30 amitk, no mentioning in dmesg and no plug events (i see its a codesourcery kernel so i'm sure its yours) Apr 09 08:48:37 * ogra checks the config Apr 09 08:49:01 ogra: try "modprobe ehci-hcd" Apr 09 08:49:14 amitk, i cant, its a server image Apr 09 08:49:24 thats why i asked if everything is compiled in Apr 09 08:49:36 * amitk prepares another kernel with ehci-hcd compiled in Apr 09 08:49:46 let me try with a live image first Apr 09 08:49:59 it just takes endless to boot compared to server Apr 09 08:50:29 and i have no SD ready with one, takes a while Apr 09 08:53:51 ogra: good idea, I'll switch to a server image too Apr 09 08:54:14 note that only works if you dont need any modules Apr 09 08:54:41 hey amitk, ogra Apr 09 08:54:49 yo lrg Apr 09 09:05:10 ogra: people now has a kernel with ehci compiled in Apr 09 09:06:19 ok Apr 09 09:11:36 i see ehci in dmesg now, but the kbd doesnt get powered on Apr 09 09:11:52 i guess hci stuff is modules Apr 09 09:12:02 * ogra tries live Apr 09 09:13:36 hci? Apr 09 09:13:51 well, input drivers Apr 09 09:16:07 aah, HID Apr 09 09:17:25 amitk, live works Apr 09 09:17:55 tell me what you will do with HID since i need to include the udeb in d-i if it stays modular Apr 09 09:18:34 hey, even my moschip USB->ETH works out of the box now Apr 09 09:19:22 * ogra fires up firefox to see if it survives with 25% compcache Apr 09 09:20:19 eggonlea: ping? Apr 09 09:20:44 www.ubuntu.com works \o/ Apr 09 09:20:50 * ogra tries something havier Apr 09 09:21:14 ogra: so EHCI needs to be compiled in for now... I plan to keep HID modular Apr 09 09:21:32 bah, hello OOM Apr 09 09:23:50 NCommander: ack Apr 09 09:27:08 the framebuffer definately has some issues with plymouth Apr 09 09:53:45 ok, compcache with 50% doesnt let firefox OOM anymore Apr 09 09:54:15 and confirmed that ehci is reqd to be compiled in Apr 09 09:54:26 will finalise the patches and prep for upload Apr 09 09:54:40 amitk, what about sound ? Apr 09 09:54:54 i dont seem to have any sound device atm Apr 09 09:55:29 morning Apr 09 09:57:04 are there plans to keep kernel+initrd in nand on omap platform? Apr 09 09:57:22 ogra: do you care about sound ATM? Apr 09 09:57:28 as you want to have rootfs on usb stick then this would drop requirement of sd card for boot Apr 09 09:57:57 amitk, well, if we can have it it would be great, not essential though Apr 09 09:58:06 amitk, what breaks it ? USB ? Apr 09 09:58:28 hrw, probably later, for 10.04 not Apr 09 09:58:34 ok Apr 09 09:58:50 way to much to do to even make the installer work in the next 7 days Apr 09 09:59:04 ogra: Hey do you know if the OMAP daily images work somewhat already? Apr 09 09:59:21 lool, they dont, we're waiting for a kernel Apr 09 09:59:29 lool, no USB support atm Apr 09 09:59:38 ogra: Do they boot without USB? Apr 09 09:59:51 e.g. if I write them to SD, will they show some UI? Apr 09 09:59:59 lool, yes, but you dont have input devices and no target to install to Apr 09 10:00:12 lool, you get all the way to the live desktop Apr 09 10:00:16 ogra: Ok thanks; What about the netboot ones? Apr 09 10:00:51 lool, i havent tested them for a while, they might work ... server definately does to a certain point if you modify boot.scr for serial Apr 09 10:01:14 (and the preseed file, i havent rebuilt since i changed debian-cd) Apr 09 10:01:49 live has a to low compcache value for installing atm Apr 09 10:01:57 ogra: Do you intend to build a SD image for d-i images? Apr 09 10:02:02 i'm pondering to use only-ubiquity Apr 09 10:02:41 lool, no, time is to short, i'll try to get server and live working ... in case there is time left i'll look into netinst d-i images like we have for imx51 Apr 09 10:02:49 but its unlikely Apr 09 10:03:43 lool, getting partman and flash-kernel hacked up to do what we need will be time consuming Apr 09 10:04:12 oh we disabled alternate desktop images for armel Apr 09 10:04:24 lool, thats what i was talking about yesterday :)# Apr 09 10:04:32 lool, server is your best bet Apr 09 10:04:53 ogra: It's disabled or simply broken? Apr 09 10:04:55 unless you want a live image Apr 09 10:04:59 ogra: Did you push the fix I suggested? Apr 09 10:05:07 lool, disabled, we dont support desktop on armel anymore Apr 09 10:05:12 lool, yup Apr 09 10:05:28 thanks for that suggestion again :) Apr 09 10:05:43 no xubuntu image anymore either hmm Apr 09 10:05:58 nope Apr 09 10:06:03 only netbook Apr 09 10:06:06 and server Apr 09 10:06:15 ogra: compiling a final kernel for test with all changes after reworking... once that is verified, I'll ask for an upload Apr 09 10:06:17 ogra, asac: Can you guys confirm that the subarch name we've been using for omap is "armel+omap"? Apr 09 10:06:22 we have massively cut down on images Apr 09 10:06:30 I see https://wiki.ubuntu.com/LucidLynx/ReleaseManifest mentions armel+OMAP3, and would like to write armel+omap there Apr 09 10:06:31 lool, confirmed Apr 09 10:06:48 amitk, yay Apr 09 10:11:40 lool, i think davidm added the 3 to indicate that we will only support omap3 by release (probably note it in brackets or something) Apr 09 10:15:04 ogra: added, thanks Apr 09 10:17:17 amitk, i see weird errors in dmesg if ubiquity tries to start partman Apr 09 10:17:28 ogra: if you'll confirm the kernel on people, we're good to upload Apr 09 10:17:42 ogra: pastebin? Apr 09 10:19:37 * ogra weaits for ubuntu-paste to return the prompt Apr 09 10:20:21 grrr, paste.ubuntu.com is broken again Apr 09 10:20:27 amitk, http://pastebin.com/khbv1c5i Apr 09 10:20:57 amitk, see the parted messages at the bottom Apr 09 10:26:34 ogra: something not using setrlimit(RLIMIT_CORE, ...) correctly? Apr 09 10:26:58 might be Apr 09 10:27:46 man 5 core Apr 09 10:30:14 ogra: it is a warning message printed by do_coredump() inside the kernel Apr 09 10:30:33 if (cprm.limit == 0) { Apr 09 10:30:33 /* Apr 09 10:30:34 * Normally core limits are irrelevant to pipes, since Apr 09 10:30:34 * we're not writing to the file system, but we use Apr 09 10:30:34 * cprm.limit of 0 here as a speacial value. Any Apr 09 10:30:35 * non-zero limit gets set to RLIM_INFINITY below, but Apr 09 10:30:36 yeah, RTLIMIT_CORE being 0 is wanted cjwatson tells me Apr 09 10:30:38 * a limit of 0 skips the dump. This is a consistent Apr 09 10:30:40 * way to catch recursive crashes. We can still crash Apr 09 10:30:43 * if the core_pattern binary sets RLIM_CORE = !0 Apr 09 10:30:45 * but it runs as root, and can do lots of stupid things Apr 09 10:30:48 * Note that we use task_tgid_vnr here to grab the pid Apr 09 10:30:50 * of the process group leader. That way we get the Apr 09 10:30:53 * right pid if a thread in a multi-threaded Apr 09 10:30:55 * core_pattern process dies. Apr 09 10:30:58 */ Apr 09 10:31:00 printk(KERN_WARNING Apr 09 10:31:03 "Process %d(%s) has RLIMIT_CORE set to 0\n", Apr 09 10:31:05 task_tgid_vnr(current), current->comm); Apr 09 10:31:08 printk(KERN_WARNING "Aborting core\n"); Apr 09 10:31:10 goto fail_unlock; Apr 09 10:31:13 cool Apr 09 10:31:21 in that case, please give the ack for the latest kernel on people Apr 09 10:31:28 ogra: ^ Apr 09 10:31:42 i'll test it, one sec Apr 09 10:32:40 https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/498525 Apr 09 10:32:44 Launchpad bug 498525 in linux (Ubuntu Lucid) (and 1 other project) "[lucid] breaks apport: core dumps get aborted even if core_pattern is a pipe (affects: 2) (heat: 16)" [High,Fix released] Apr 09 10:34:18 amitk, so that fix is missing from the omap kernel ? Apr 09 10:34:24 apw: this bug above ^^ Apr 09 10:34:31 ogra: I think so Apr 09 10:34:49 apw: I guess the patch you created should be applied to all flavours? Apr 09 10:35:06 amitk, it is allplied to all flavours based on .32 Apr 09 10:35:22 its that .33 was not taken from leaans rebase means you didn't get it Apr 09 10:35:30 apw: I'm going to try cherry-picking it into omap Apr 09 10:35:37 amitk, ok Apr 09 10:35:48 we likely need to put that on the release sprint agenda too Apr 09 10:35:55 deciding what to do about .33 Apr 09 10:36:20 apw, we have no committed support for omap at all Apr 09 10:36:25 personally i think it should be synced with leaans pass though .33 version. she was said to have kept it Apr 09 10:37:30 apw, https://wiki.ubuntu.com/LucidLynx/ReleaseManifest Apr 09 10:38:18 i wonder what N/A means in that context Apr 09 10:38:19 apw, its pretty much "get something out that runs and install roughly and dont care anymore" Apr 09 10:38:36 but regardless of what you say we will be expected to do secirity on it Apr 09 10:38:42 which means it doesn't want to be an unmaintainable mess Apr 09 10:38:44 we need to have an installable image by release but dont do any further stuff with it Apr 09 10:39:11 i will believe that when we get to M without updating it Apr 09 10:49:50 ogra: feedback? Apr 09 10:52:10 amitk, installing ... Apr 09 10:55:58 crap, OOmed when trying to install it on the running system Apr 09 11:03:35 ogra: 364MiB is the lower bound. How much are you running with omap? Apr 09 11:03:45 ogra: ^we ever successed ran with on Dove Apr 09 11:04:15 NCommander, i know the minimal values Apr 09 11:04:25 256M plus 50% compcahce atm Apr 09 11:04:48 ogra: hrm, that should push you just enough to complete installation. Doesn't ubiquity activate swap space like d-i does? Apr 09 11:23:50 1 Apr 09 11:23:52 Ups Apr 09 11:27:43 sigh, only-ubiquity dies too in ubi-partman Apr 09 11:38:47 ogra: OOM got it? Apr 09 11:38:52 or something else? Apr 09 11:43:22 ubi-partman Apr 09 11:43:25 no OOM Apr 09 11:44:33 ogra: oh. I though ubi-partman was dying due to OOM before Apr 09 11:46:42 no, it does with "error 10" Apr 09 11:47:00 i suspect its caused by udisks-part-id failing Apr 09 11:48:02 ogra: why is that failing Apr 09 11:48:08 no idea yet Apr 09 11:48:45 * NCommander decides this is a good time to fix apport Apr 09 11:48:57 apport works fine for me Apr 09 11:49:13 ogra: the retracer been down for two weeks, chroot broke Apr 09 11:49:14 i just filed bug 559144 with it Apr 09 11:49:15 Launchpad bug 559144 in udisks (Ubuntu Lucid) (and 1 other project) "udisks-part-id crashed with signal 7 in memcpy() (affects: 1)" [High,New] https://launchpad.net/bugs/559144 Apr 09 11:49:26 and I couldn't rebuild the chroot on jocote Apr 09 11:49:41 * NCommander points to apport stauts in the topic Apr 09 11:52:17 err Apr 09 11:52:23 * ogra glares at the bugreport Apr 09 11:52:27 ProcCmdLine: quiet splash vram=12M omapfb.mode=dvi:1280x720MR-16@60 file=/cdrom/preseed/hostname.seed -- boot=casper Apr 09 11:52:36 wtf Apr 09 11:53:14 hostname.seed? Apr 09 12:04:58 must be something that parses the line its surely not in the actual cmdline Apr 09 12:08:12 msg ndechesne comment ca va? Apr 09 12:08:31 heh Apr 09 12:08:38 * ogra hands OlivierP a / Apr 09 12:14:47 ogra: Just checking lucid-netbook-armel+omap.img, apparently it doesn't follow the chs expectations for beagleboard? Apr 09 12:14:57 Partition 1 has different physical/logical endings: phys=(1023, 3, 32) logical=(8283, 2, 1) Apr 09 12:15:05 it woprks so who cares Apr 09 12:15:42 it just uses the dove scripts in slightly modified ways, CHS is only relevant if you use MLO on the image which we dont do Apr 09 12:16:01 ogra: Do you have u-boot on the image? Apr 09 12:16:07 lool, nopüe Apr 09 12:16:15 erks, I have no MLO on any of my boards Apr 09 12:16:24 the expectation is that x-loader and u-boot sit in NAND Apr 09 12:16:39 ogra: So the image only works if people have a working x-loader and u-boot in NAND with proper config? Apr 09 12:16:44 Ok Apr 09 12:16:48 right Apr 09 12:16:59 for 10.04 thats the most minimal Apr 09 12:17:04 I thought you were trying to change the image scripts to use proper partitions Apr 09 12:17:11 for later we'll come up with something better Apr 09 12:17:21 lool, i will, but not for 10.04 Apr 09 12:17:59 we're way to late, my target is to have something installable at all by release going the path of least resistance Apr 09 12:19:52 you guys do have the script to make correct partitions? Apr 09 12:21:19 XorA: No; only the public (manual) instructions Apr 09 12:21:57 XorA: unfortunately, our scripts are usually based on parted which has a rigid view of what values one should use for chs addressing Apr 09 12:21:58 lool: http://git.openembedded.net/cgit.cgi/openembedded/tree/contrib/angstrom/omap3-mkcard.sh Apr 09 12:22:10 lool: for reference Apr 09 12:23:03 XorA: Thanks; we considered using other disk utilities such as fdisk or sfdisk or others, but that's an helpful example Apr 09 12:23:28 lool: I still get 50 odd downloads of that a day from new beagleboard owners Apr 09 12:23:43 lool, i'll need to have partman-uboot, flash-kernel etc changes before final freeze (and whatever else bugs we find during next week) thats work enough given we only have a usable kernel today Apr 09 12:23:46 * ogra sighs about reconnects Apr 09 12:24:16 i'll dd the working live image off my Sd and upload it to people.u.c so others can play as well Apr 09 12:24:16 XorA: it might make sense to round the size down to the nearest cylinder boundary before dd-ing zeroes Apr 09 12:24:56 lool: that makes sense Apr 09 12:25:00 XorA: For our particular case, I wish we wouldn't need root (we currently don't have root on the image build server) Apr 09 12:25:48 XorA: Also, I'm afraid the scripts relies on using real hardware of type MMC to create the image?! Apr 09 12:25:58 It seems to rely on a /dev/foop2 device for instance Apr 09 12:26:14 could you repost the script ? i was disconnected Apr 09 12:26:18 you can do it be prepping a blank image in a file once, then writing your FS on top of that Apr 09 12:26:24 http://git.openembedded.net/cgit.cgi/openembedded/tree/contrib/angstrom/omap3-mkcard.sh Apr 09 12:26:25 That only exists for real devices (unless you ask kpartx) and not for USB attached MMC drives Apr 09 12:26:33 thanks :) Apr 09 12:27:19 ah, yeah, i had a similar approach using sfdisk Apr 09 12:30:39 * ogra uploads http://people.canonical.com/~ogra/lucid-netbook-armel+omap.img Apr 09 12:30:59 that uses 50% compcache and has working USB Apr 09 12:31:09 awesome Apr 09 12:31:14 40min to go Apr 09 12:31:32 ogra: on what? Apr 09 12:31:38 NCommander, ?? Apr 09 12:31:39 * NCommander does his first likewise-open build Apr 09 12:31:48 ogra: 40 mins to go on what? Apr 09 12:31:51 ECC Failed, page 0x00080000 Apr 09 12:31:52 gah Apr 09 12:31:55 NCommander, the upload Apr 09 12:32:03 lool, wait for the image above Apr 09 12:32:15 lool, the cdimage images all have the broken kernel Apr 09 12:32:19 ogra: ugh, your upload bandwidth supassing mine in spades; with my up, that would take about 4 hours Apr 09 12:32:33 NCommander, SDSL 2Mbit Apr 09 12:32:56 ogra: I have 15 down burst up to 50Mbit, with 1Mbit up Apr 09 12:32:57 while it sucks wrt downloading i have a guranteed 2M upload all the time Apr 09 12:32:58 ;.; Apr 09 12:33:03 and the 1 is questionable Apr 09 12:33:05 ogra: I'll trade :-P Apr 09 12:33:18 NCommander, costs a fortune though Apr 09 12:33:37 * XorA has that ADSL where I can request to boost my upload at cost of download Apr 09 12:33:41 ogra: can't be as bad as what I'm paying to get 1 Mbit up versus 264k Apr 09 12:33:47 XorA: I envy you Apr 09 12:33:57 XorA, germany doesnt have such offers sadly Apr 09 12:34:12 ogra: its rare in the UK, I think this ISP is the only one Apr 09 12:34:16 XorA: nor do most ISPs int he states Apr 09 12:34:38 I havent here, but my office does have it boosted Apr 09 12:36:02 ogra: This is before the kernel Apr 09 12:36:13 lool, urgh Apr 09 12:36:23 bad u-boot ? Apr 09 12:37:17 wrote u-boot with the wrong mixed up sw/hw ecc settings I would guess Apr 09 12:38:03 XorA, yeah, lool doesnt trust me so he doesnt use my packages :P Apr 09 12:38:20 we have MLO and u-boot.bin in the archive :) Apr 09 12:38:38 ogra: I'm using MLO and u-boot.bin from the archive... Apr 09 12:38:39 * XorA hasnt ubuntud any of his beagles yet Apr 09 12:38:52 lool, aw, C4 ? Apr 09 12:38:58 qemu-maemo-system-arm Apr 09 12:39:03 ah Apr 09 12:39:05 phew Apr 09 12:39:24 i tested it on C4 Apr 09 12:42:09 lool, qemu-maemo-system-arm might emulate a different revision in which case you probably need a different MLO Apr 09 12:43:33 * ogra considers to be brave and try what the live image does on his B6 Apr 09 12:45:41 ogra: I just managed to load our u-boot, but can't break into it Apr 09 12:45:48 ogra: er, doesn't your B6 only have 128M? Apr 09 12:45:58 NCommander, thus "brave" :) Apr 09 12:46:18 ogra: ah. Ping me when its done booting sometime after final freeze ;-) Apr 09 12:46:22 seems though our DSS2 already fails on it i get no graphical output Apr 09 12:46:42 its stuck on the u-boot splash Apr 09 12:46:51 Is there anyway to break the start of our u-boot? Apr 09 12:47:06 lool: it should break on any key Apr 09 12:47:08 lool, only hitting a key Apr 09 12:47:15 while the countdown runs Apr 09 12:47:41 I don't see any coundtown Apr 09 12:47:56 are you sure its our u-boot ? Apr 09 12:47:57 Note that it uses the default environment; I don't have any default binary environment Apr 09 12:48:04 U-Boot 2010.03-rc1 (Mar 24 2010 - 15:50:56) Apr 09 12:48:10 it definately defaults to a 10 sec countdown Apr 09 12:48:11 ogra: I don't see which other one it could possibly be Apr 09 12:48:12 wow Apr 09 12:48:15 its up ot date Apr 09 12:48:21 * NCommander is sad being on u-boot 1.4.3 ;.; Apr 09 12:48:34 what I get is http://paste.ubuntu.com/411595/ Apr 09 12:48:35 NCommander, yes, i pulled the very latest upstream that upstream recommended to me Apr 09 12:49:02 ogra: Dove's u-boot is a dinosaur. Although its not as bad as things still shipping u-boot 1.1, and Marvell has added some nice features Apr 09 12:49:07 lool, seems it'S stuck, the countdown appears after that Apr 09 12:49:10 like mounting NFS in u-boot :-) Apr 09 12:49:49 lool, http://paste.ubuntu.com/411596/ Apr 09 12:50:32 ah yes, ran into that in real hardware, if you us angstrom's u-boot 2010.03-rc1 you have to use the new 1.4.4ss x-loader... Apr 09 12:50:45 Start in 49 minutes (5000000) - wow, buildds are getting slammed this morning Apr 09 12:51:26 * ogra hopes thats not the omap kernel Apr 09 12:51:27 ogra: that assumes you have some delay Apr 09 12:51:36 rcn-ee: Ah Apr 09 12:51:40 ogra: no, that's compiz Apr 09 12:51:46 rcn-ee: Do you have a binary x-loader I could try out to confirm? Apr 09 12:51:47 ah, k Apr 09 12:51:51 rcn-ee: Preferably signed Apr 09 12:51:54 yeap.. Apr 09 12:51:55 ogra: ti kernel is building, I set it so high that it was picked up as soon as a buildd freed up Apr 09 12:52:03 NCommander, yup, i see it Apr 09 12:52:22 lool, they are uploaded here http://rcn-ee.net/deb/tools/ Apr 09 12:52:23 ogra: why are you hoping it isn't the kernel? Apr 09 12:52:31 * NCommander watchs his racing OOo builds Apr 09 12:52:35 amitk, ?? Apr 09 12:52:49 amitk, i said i suspect it *is* the kernel Apr 09 12:52:57 15:51 * ogra hopes thats not the omap kernel Apr 09 12:53:06 amitk: he was worried that the 50 minute build delay was for the kernel Apr 09 12:53:07 ah Apr 09 12:53:14 i was referring to the 49min delay Apr 09 12:53:30 * NCommander looks at the build queue to shuffle things around Apr 09 12:53:30 amitk, i thoght you referred to the SIGBUS errors Apr 09 12:53:33 Got to love crunch time Apr 09 12:53:40 so heads up, that'll be in the FAQ's.. on boards with an older x-loader (1.4.2 is on C4's) when you upgrade u-boot to 2010.03, you need to flash the new 1.4.4ss X-loader... (or hold the user botton down on reset with the MLO on the sd card) Apr 09 12:54:16 otherwise it'll just stall at u-boot load.... Apr 09 12:54:21 * ogra twiddles thumbs ... Apr 09 12:54:25 rcn-ee: Your x-loader is unhappy with my qemu Apr 09 12:54:25 Reading boot sector Apr 09 12:54:25 Error: reading boot sector Apr 09 12:54:25 u-boot.bin not found or blank nand contents - attempting serial boot . . . Apr 09 12:54:29 another 15min until the image is uploaded Apr 09 12:54:32 Texas Instruments X-Loader 1.4.4ss (Apr 1 2010 - 07:01:03) Apr 09 12:54:32 Beagle xM Rev A Apr 09 12:54:41 rcn-ee: it waits for an u-boot over kermit Apr 09 12:54:47 XM ? Apr 09 12:55:07 * amitk starts debugging OTG and kernel boot on XM now Apr 09 12:55:10 ah crap... it's build directly in angstrom... Apr 09 12:55:15 * ogra really doubts qemu can emulate XM already Apr 09 12:56:10 amitk, kernel doesnt boot on revB Apr 09 12:56:12 :) Apr 09 12:56:20 (not that i expected it to work) Apr 09 12:56:21 lool, the last version that worked before the 1.4.4ss change, was 2009.11 (in the same directory) Apr 09 12:57:06 ogra: I don't have a revB. Only C4 and XM Apr 09 12:57:17 amitk, lucky you ... revB is crap Apr 09 12:57:33 128M simply dont cut it :) Apr 09 12:58:08 rcn-ee: Wee! Apr 09 12:58:10 laughs, it's sad I use Rev b's for all my testing because my C's and Xm's are doing all the real grunt work on the server.. ;) Apr 09 12:58:26 rcn-ee: So I understand that we effectively packaged conflicting x-loader + u-boot in Ubuntu? Apr 09 12:58:27 it is still good for lots of things, like running a server, but I'll need someone else to debug the problem Apr 09 12:58:41 lool, huh ? Apr 09 12:59:01 lool, we packaged what sarkoman recommended Apr 09 12:59:02 * NCommander pedals faster Apr 09 12:59:13 lool, for revC boards Apr 09 12:59:13 rcn-ee: want to see what is preventing the lastest kernel upload from booting on revB? Apr 09 12:59:19 ogra: our x-loader + our u-boot == fail, our x-loader + rcn-ee's older u-boot == work Apr 09 12:59:26 lool, no the x-loader and u-boot in ubuntu has all the XM tweaks, i think qemu might be missing one of the new changes... Apr 09 12:59:28 lool, in qemu Apr 09 12:59:34 rcn-ee: Ok Apr 09 12:59:38 rcn-ee, it doesnt Apr 09 13:00:06 amitk, will do, i took half of today off, so starting noon (central) i'm free for regression stuff. Apr 09 13:00:07 rcn-ee, XM support landed after my checkout, i asked sarkoman explicitly for the best combo for revC Apr 09 13:00:29 at least I can get qemu as far as loading the kernel and uncompressing it Apr 09 13:00:32 amitk, even if the kernel boots it wont be usable anyway Apr 09 13:00:40 I'm afraid nothing happens afterward Apr 09 13:01:03 rcn-ee: thanks! people.canonical.com:~/amitk/ti/ has the .deb if you don't want to wait for the official builds Apr 09 13:01:15 ogra, i guess i haven't looked too closely at which checkout ubuntu pulled... Apr 09 13:01:41 ogra: why can't you pull an update of xloader and uboot? Apr 09 13:02:03 amitk, because i'm busy debugging your kernel and the SIGBUS ? Apr 09 13:02:20 amitk, i can pull an update on the weekend Apr 09 13:02:41 aah, I thought there was some more sinister reason Apr 09 13:02:43 amitk, but since we wont support XM for 10.04 thats moot, i dont want to break Cx support Apr 09 13:04:09 are you guys planning a 10.04.1 arm rebuild? (i'm hoping to sneak some xm patches for that.. with amitk's blessing of course. ;)) Apr 09 13:04:17 rcn-ee, nope Apr 09 13:04:29 10.04 will be a one shot omap3 revC image Apr 09 13:04:44 rcn-ee: we can always do PPA if these mean devs don't let us push stuff into the archive ;) Apr 09 13:05:05 right, PPA is fine Apr 09 13:05:16 rcn-ee: Do feel free to push as many interesting patches on launchpad Apr 09 13:05:17 okay good to know... and i can always write a script to tweak the image.... ;) Apr 09 13:05:37 for 10.04 the plan is to have *something* Apr 09 13:05:45 which we are still far away from Apr 09 13:06:45 it's good ground work for 10.10... More of the omap stuff should be upstream by then, including the multi omap stuff.. Apr 09 13:06:53 rcn-ee: I'm curious though if you were using your EHCI port for the USB bug? Or do you plug in your hub into the OTG port? Apr 09 13:07:38 For that bug testing, I'm just using a Bx board. so only OTG (otg to usbA adapter and hub..) Apr 09 13:08:50 WOOO, OOO EXPLODED! Apr 09 13:08:51 yay Apr 09 13:08:55 (I think) Apr 09 13:09:20 rcn-ee: aah, the revBs have no EHCI port soldered onto the pad, I believe... Apr 09 13:09:49 (broken ehci implementation and all that) Apr 09 13:09:53 EVERYONE: http://people.canonical.com/~ogra/lucid-netbook-armel+omap.img ********* please help debugging this image !!! ************** Apr 09 13:10:06 should properly work on Cx boards Apr 09 13:10:10 ogra: do we have a guide for doing omap on QEMU? Apr 09 13:10:22 Correct... however the EHCI still pops up in lsusb, it's just not soldered to anything, since that port was snafu on the omap35x. the Cx boards route to a different pin.. Apr 09 13:10:25 NCommander, ask lool, he does the qemu-maemo playing atm Apr 09 13:10:41 * XorA grabs ogras image Apr 09 13:10:45 :D Apr 09 13:10:58 does it have zippy patches? Apr 09 13:11:01 you'll get into a live session properly Apr 09 13:11:22 it has whatever will end up in the archive tonight Apr 09 13:11:39 plus using 50% compcache to prevent OOM Apr 09 13:11:51 XorA, last i tried you have to be careful with the zippy patches, as it causes boards to lock up post 2.6.29... (but i haven't testd my zippy2 witk Koen's new u-boot that reads the i2c) Apr 09 13:12:03 i.e. you can run firefosx in the live session without running out of ram Apr 09 13:12:26 i dont think amitk added zippy patches Apr 09 13:12:26 (zippy2 patch kernel + board with no zippy2 = lockup) Apr 09 13:12:36 bang goes me network :-) Apr 09 13:12:48 XorA: no zippy patches Apr 09 13:13:05 XorA, do you have a zippy1 or zippy2? I have pre-build lucid chroot *.deb on rcn-ee.net for zippy2.. Apr 09 13:13:13 my USB->ETH adapter works fine though Apr 09 13:13:38 rcn-ee: zippy1, zippy2 wasnt public when I met prpplague Apr 09 13:15:01 *heads off to work.. Apr 09 13:16:00 nosse1: you could test out the above image too; I've disabled sound on the EVM boards for now. So if you can tell me if this allows you to boot, that would be great. Apr 09 13:26:35 Does anyone know how the apport retracing service works? Currently it doesn't generate useful output because gdb needs debug symbols in order to make a reasonable attempt at the backtrace. Apr 09 13:28:57 dmart: We should be providing these debug symbols in the form of -dbgsym packages Apr 09 13:29:08 wohooo !!! Apr 09 13:29:13 my zoom2 arrived Apr 09 13:29:56 lool: yes, but retracing service doesn't seem to have them installed when it generates backtraces Apr 09 13:30:10 ogra: very 80s look Apr 09 13:30:13 ogra: zoom2 or 3? Apr 09 13:30:22 lots of black plastic Apr 09 13:30:30 ;) Apr 09 13:30:51 XorA, zoom2 with upgrade kit Apr 09 13:30:56 ogra: bah Apr 09 13:31:02 dmart, giant blackberry Apr 09 13:31:03 ogra: Ive been waiting weeks for mine Apr 09 13:32:28 dmart: I think you can try replicating the apport behavior locally with apport-retrace Apr 09 13:32:42 dmart: the ddebs should be fetched from ddebs.ubuntu.cim Apr 09 13:33:03 Is apport-retrace supposed to do that automatically? Apr 09 13:33:07 Yes Apr 09 13:33:11 I think :) Apr 09 13:34:28 Also, I find the name of the debug package is unpredictable, because some package generate their own -dbg in the main archive, others generate -dbgsym on ddebs.ubuntu.com, and there seems not always to be a 1:1 relationship between binary packages and debug packages Apr 09 13:34:39 Is there an easy way to guess the correct debug packages? Apr 09 13:36:19 Unfortunately not. Apr 09 13:36:22 dmart: Ignore -dbg packages Apr 09 13:36:34 dmart: We only care about -dbgsym ones which should be generated almost all the time Apr 09 13:36:40 dmart: We just inherit -dbg from Debian Apr 09 13:37:08 Um, not quite. We have some -dbg native to Ubuntu, and packages with nostrip don't generate -dbgsyms Apr 09 13:38:43 persia: I'm not sure what you mean Apr 09 13:39:40 1) There exist packages in the Ubuntu archive that have -dbg packages that do not have -dbg packages in Debian. 2) Packages that don't call dh_strip in the usual fashion don't get dbgsyms: this is not uncommon for packages that also have -dbg variants. Apr 09 13:39:54 dmart: it does Apr 09 13:40:12 dmart: the retracer isn't very happy this morning, and the image may have had stale debs installed Apr 09 13:40:13 apport-re Apr 09 13:40:37 When the apport retracing service appends backtraces to bug reports they tend to look like this: http://pastebin.ubuntu.com/411615/ Apr 09 13:40:46 I was hoping we might be able to improve it... Apr 09 13:41:10 dmart: I'm aware, but if the dbgsym's are out of date/newer than what is available, the retrace explodes. Also some kernel bugs can cause crappy retraces Apr 09 13:41:35 dmart: I'm working on the broken OOo segfault, I think we're looking at a toolchain regression, bisecting now Apr 09 13:41:38 Well I can't go past Uncompressing Linux... with any kernel I try, hmm Apr 09 13:45:13 Do we not keep old debug packages on the servers for a bit? Apr 09 13:45:44 dmart: Not too much I'm afraid Apr 09 13:47:08 persia: Concerning -dbg which we add in Ubuntu, we still have -dbgsym for them; we only provide -dbg for end-users, not apport, right? and concerning packages not using dh_strip and not having -dbgsym as a result, these are bugs/limitations, but the apport model still relies on -dbgsym AFAIK Apr 09 13:47:50 dmart: /win 31 Apr 09 13:47:52 er Apr 09 13:47:58 The vast majority of packages with ELF binaries have -dbgsym packages and that's the main use case I believe Apr 09 13:48:04 NCommander: /win 42 to you too Apr 09 13:49:43 lool: heh, I've jumped from x-chat to irssi Apr 09 13:49:57 lool: I'll agree with you that apport relies on -dbgsym. I just want to make sure everyone understands that it's not quite perfect yet. Apr 09 13:50:32 lool: sure, but many key projects have their own -dbg packages instread, so it's hard to automate fetching the debug packages. Apr 09 13:51:24 dmart: As I said, the machine-usable service is based around -dbgsym packages -- not -dbg; we can fix these if they are missing for this or that package/architecture Apr 09 13:51:43 Forcibly pulling in libc6-dbg might help Apr 09 13:51:43 dmart: or what do you mean with "key projects"? Apr 09 13:52:08 Things like eglibc, firefox etc. Apr 09 13:52:25 dmart: Well you typically want libc6-dbgsym_2.11.1-0ubuntu5_armel.ddeb Apr 09 13:52:35 hmmm Apr 09 13:52:44 * dmart wonders why he failed to find that in the past... Apr 09 13:53:00 dmart: You need to add the ddeb repo to your sources.list and then call apport-retrace I think Apr 09 13:53:05 (and apt-get update inbetween) Apr 09 13:53:19 dmart: -dbg are not meant for apport, but for end-user or specific use cases (e.g. valgrind) Apr 09 13:53:36 Does apport-retrace pull in the debug packages for the whole library stack? This is the other problem we get: the backtrace will tend to die whenever it reached a frame with no debug info. Apr 09 13:53:36 or to help Debian-style debugging Apr 09 13:53:56 dmart: In theory it should do something ensuring it gets the proper -dbgsym packages installed Apr 09 13:55:30 Do you know whether old debug packages are mirrored / backed up somewhere? Maybe there is a problem with them often having disappeared by the time a bug gets reported. Apr 09 13:55:30 dmart: there's an apport-retracer chroot creation script which comes at least in the apport source Apr 09 13:55:56 Hmmm, I'll try and have a play with that at some point Apr 09 13:55:58 dmart: Yes, that's common indeed; I'm afraid we don't keep them for too long as this is huge; pitti might have some details there Apr 09 13:57:01 I think some work is being done to improve gdb's backracing, but I don't think it's all merged yet, unfortunately... Apr 09 13:57:27 dmart: ack; -dbgsym are meant exactly for what you need in the mean time I guess Apr 09 13:57:46 ok, I'll try to persevere with that Apr 09 13:58:22 NCommander: it sounds like you might need to grab some dbgsyms if you want a better backtrace for the bash segfault in the OOo build Apr 09 13:58:29 dmart: Basically, we divert dh_strip during package builds and take a copy of the debug symbols just before calling the real dh_strip Apr 09 13:58:46 because it's made automatically for all build, it covers almost all packages Apr 09 13:59:46 Maybe the difficulties I had previously were to do with disappeared package versions. It was a while ago now. Apr 09 14:05:13 NCommander: Am I understanding the OOo build issue correctly? It looks to me like a bash crash in the build and not something directly connected with the OOo sources? Apr 09 14:09:51 dmart: its not a bash crash Apr 09 14:12:31 Is this bug 555977? Apr 09 14:12:33 Launchpad bug 555977 in openoffice.org (Ubuntu Lucid) (and 1 other project) "openoffice.org FTBFS on armel (affects: 1) (heat: 6)" [High,Triaged] https://launchpad.net/bugs/555977 Apr 09 14:13:15 dmart: thats it Apr 09 14:13:15 ndechesne, http://people.canonical.com/~ogra/lucid-netbook-armel+omap.img and http://people.canonical.com/~ogra/live-initrd-roller.sh (note that your kernel needs squashfs and aufs at least) Apr 09 14:13:39 dmart: I'm confirming if we're looking at regression in gcc. If it is, at least its a lot less of a nightmare to debug it Apr 09 14:14:29 What's the nature of the problem exactly? I thought it was a build failure: Apr 09 14:14:32 /bin/bash: line 1: 24891 Segmentation fault Apr 09 14:14:59 dmart: it is, due to saxparser segfaulting, which appears to be a regression caused by a toolchain update Apr 09 14:15:08 Oh, right--- my confusion Apr 09 14:15:46 * NCommander pedals faster Apr 09 14:56:47 bug 559295, bug 559301 and bug 559297 for omap issues Apr 09 14:56:52 Launchpad bug 559295 in flash-kernel (Ubuntu Lucid) (and 1 other project) "flash-kernel-installer needs to learn to handle omap (affects: 1)" [High,New] https://launchpad.net/bugs/559295 Apr 09 14:56:53 Launchpad bug 559301 in partman-uboot (Ubuntu Lucid) (and 1 other project) "partman-uboot needs to handle omap installs (affects: 1)" [High,New] https://launchpad.net/bugs/559301 Apr 09 14:56:55 Launchpad bug 559297 in flash-kernel (Ubuntu Lucid) (and 1 other project) "flash-kernel needs to learn to handle omap (affects: 1)" [High,New] https://launchpad.net/bugs/559297 Apr 09 14:58:06 bug 559144 Apr 09 14:58:07 Launchpad bug 559144 in udisks (Ubuntu Lucid) (and 1 other project) "[armel] udisks-part-id crashed with signal 7 in memcpy() (affects: 1) (heat: 12)" [High,New] https://launchpad.net/bugs/559144 Apr 09 15:03:12 bug 541399 Apr 09 15:03:14 Launchpad bug 541399 in debian-installer (Ubuntu) "netboot image fails to boot. (affects: 1) (heat: 8)" [Medium,Fix committed] https://launchpad.net/bugs/541399 Apr 09 15:10:08 ogra: thanks. will try this. Apr 09 15:12:27 ndechesne: i asked cpearson something in pmsg ;) Apr 09 15:12:34 NCommander: can you poke him to look at his client ;) Apr 09 15:12:42 ndechesne: oh sorry Apr 09 15:12:51 ndechesne: figured that you are not on the same continent ;) Apr 09 15:12:54 nev3rmind Apr 09 15:14:34 hahaha Apr 09 15:14:49 ndechesne, just stretch your arm ! Apr 09 15:16:18 asac: bug #541030 is omap related for your list, and just went to fix committed Apr 09 15:16:21 Launchpad bug 541030 in linux-ti-omap (Ubuntu Lucid) (and 1 other project) "omap kernel musb/ehci ports not enabled on beagleboard (affects: 1) (heat: 14)" [High,Fix committed] https://launchpad.net/bugs/541030 Apr 09 15:17:06 plars: hmm ogra said thats fixed ... so didnt mention it at all Apr 09 15:17:07 let me add Apr 09 15:17:16 asac: also, bug #542041 was dropped, unless it becomes reproducible again (it was from before we had images) Apr 09 15:17:17 Launchpad bug 542041 in linux-ti-omap (Ubuntu Lucid) (and 1 other project) "ext4 support broken on omap kernel (affects: 2)" [High,Invalid] https://launchpad.net/bugs/542041 Apr 09 15:17:53 asac: hmm... he probably knows better than I, I didn't go look to see if it had already been uploaded. If it's not fixed, then it's *just* about to be Apr 09 15:18:06 plars, it was also a different kernel Apr 09 15:18:14 right Apr 09 15:18:15 leave the ext4 one out for now Apr 09 15:18:16 plars: yeah. now its fix committed in the report ;) Apr 09 15:18:30 ogra: so we should untarget the ext4 bug Apr 09 15:18:45 ogra: # Apr 09 15:18:45 559295: flash-kernel-installer needs to learn to handle omap Apr 09 15:18:45 # Apr 09 15:18:46 559297: flash-kernel needs to learn to handle omap Apr 09 15:18:49 shouldnt that be one bug ;)? Apr 09 15:18:55 done Apr 09 15:18:55 no, its two apps Apr 09 15:19:06 the untarget that is Apr 09 15:19:06 ok Apr 09 15:19:07 i want the two implementation details distinct Apr 09 15:19:11 thanks plars Apr 09 15:19:18 its the same package though Apr 09 15:19:35 ogra: then its one bug ;) Apr 09 15:19:41 sourc epackage needs to learn ompap Apr 09 15:19:43 but ok Apr 09 15:19:46 we will survive i am sure Apr 09 15:20:01 well, one part affects the installer only, the other the general upgradeability Apr 09 15:20:24 their possibility of impact is different Apr 09 15:20:38 and i will neeed FFe bugs for both features i guess Apr 09 15:26:06 bug 541399 Apr 09 15:26:09 Launchpad bug 541399 in debian-installer (Ubuntu) "netboot image fails to boot. (affects: 1) (heat: 8)" [Medium,Fix committed] https://launchpad.net/bugs/541399 Apr 09 15:26:31 ogra: ping Apr 09 15:26:39 prpplague, yep Apr 09 16:21:50 amtik, I'm testing the image now. However, I don't think support for the specific LCD panel of the EVM has been compiled into the kernel, so the panel doesn't work. Apr 09 16:22:29 Is there a boot parameter for making the installer run solely in text mode (against console) ? Apr 09 16:23:21 nosse1, no, its a live image Apr 09 16:23:49 identical to UNR on i386 Apr 09 16:24:29 nosse1: remove "quiet splash" to atleast be able to see the kernel bootup. And add "console=.... serialtty=ttyS2" to enable serial console in your boot.cmd Apr 09 16:24:38 ogra_cmpc, ok. Is it possible to run the _live image_ without a gfx scereen? Apr 09 16:24:45 and then compile it to a uboot script format Apr 09 16:24:53 amitk, done that. The last thing the console displays is: Apr 09 16:24:57 no, you can have a serial login though Apr 09 16:26:52 Here's the output I got: http://pastebin.com/Lii2QqZx Apr 09 16:28:21 you are missing serialtty=ttyS2 in your cmdline Apr 09 16:29:32 ogra_cmpc, No, I didn't. But I did initially. Thats why there are two lines with setenv bootargs Apr 09 16:30:01 note serialtty != console Apr 09 16:30:12 ah, sorry Apr 09 16:30:18 * nosse1 retrying Apr 09 16:30:20 you need serialtty=ttyS2 additionally Apr 09 16:30:54 that will tell casper to spawn a login shell on the serial tty oyu defined Apr 09 16:33:05 amitk, hmm, i guess with adding the DSS2 stuff the kernel doesnt work on zoom anymore Apr 09 16:33:20 i dont get it to uncompress Apr 09 16:34:52 Sorry, It stops at the same place as the pastebin above. Apr 09 16:36:07 can you pastebin it again ? Apr 09 16:36:11 I should note that I have not been able to run Ubuntu yet on target, even with rootstock images Apr 09 16:36:15 Yes, sure. Hold on Apr 09 16:37:14 http://pastebin.com/WK9CNB2M Apr 09 16:38:53 and disk access stops at that point ? Apr 09 16:39:43 ogra_cmpc: uncompressin almost never has anything to do with drivers Apr 09 16:40:08 Yes it seems so. I don't really know, as there's no led on the sdcard activity. Apr 09 16:40:32 I *could* bring out my scope Apr 09 16:40:48 amitk, hmm, it saw it booting with the old archive kenrel in nice Apr 09 16:43:34 amtik, How long time does it take to until the new omap kernel makes it to ports.ubuntu.com ? Apr 09 16:43:50 https://edge.launchpad.net/ubuntu/+source/linux-ti-omap/2.6.33-500.5/+build/1684225 Apr 09 16:43:54 still building Apr 09 16:44:19 excellent Apr 09 16:44:38 once thats done an archive admin needs to approve it to move it out of the NEW queue Apr 09 16:44:51 When its done, I'll give it another go on the AM3517 EVM Apr 09 16:44:52 then apw needs to upload a new meta Apr 09 16:45:25 ogra_cmpc, shouldn't need a meta for ti-omap, no abi-bump Apr 09 16:45:25 nosse1, the kernel just building is identical to the one in the image you just tried Apr 09 16:45:34 apw, ah, sweet Apr 09 16:45:57 hmm, no abi bump should also mean no NEWing Apr 09 16:46:02 once you have a good kernel which boots in an image, then we'll bump so you keep that Apr 09 16:46:07 ogra_cmpc, indeed it should Apr 09 16:46:09 I just re-installed the Dove board, and the tegra2, and the smooth-stone prototype board, and the versatile express 2 with the latest lucid build .. looks good Apr 09 16:46:13 sweet Apr 09 16:46:42 ogra_cmpc, Ah, so I can take the uImage and extract the kernel/modules from the rootfs. I'll try Apr 09 16:46:45 tasksel server takes FOR-frigging-ever to run on the versatile express 2 Apr 09 16:46:46 ogra_cmpc, in theory its a no-watch situation, of course pkgmgler will core dump and we'll have to put it back Apr 09 16:46:48 Martyn, so you like the new themes ? :P Apr 09 16:46:57 ogra : I'm ... ambivalent Apr 09 16:47:10 ogra : I've gotten _so_ used to the old brown theme, that the new theme is .. disturbing Apr 09 16:47:22 * ogra_cmpc likes it Apr 09 16:47:29 the brown got tiring Apr 09 16:47:46 and i have seem whats ahead the road for M :) Apr 09 16:48:12 changing the color is only the first step towards revolution :) Apr 09 16:48:15 right now though, I'm less worried about THEME as getting server -stable- on lucid Apr 09 16:48:40 whats unstable about it ? Apr 09 16:48:41 we're missing some mission-critical libraries, and there's nothing that can be done for Lucid Apr 09 16:49:00 there will be no way to compile hiphop on the ARM platform until libtbb2 is ported and tested Apr 09 16:49:11 memcached is unstable, but useable Apr 09 16:49:30 ah, yeah, armel PPAs would really be a good thing Apr 09 16:49:59 latest mysql runs, but has performance issues even when the proccessor is quad core and there are 4MB of L2 cache. No idea where the slowness is coming from, but doing the streams benchmark is giving me an idea Apr 09 16:50:00 so you could easily build your own stuff on top of the distro Apr 09 16:50:19 ogra : Yeah .. well .. you know my goal is to make M //the// distribution for server Apr 09 16:50:22 on armel Apr 09 16:50:24 Martyn, indeed you filed a huge pile of bugs for all these issues :) Apr 09 16:50:34 ogra : no, I //will// be filing them for M Apr 09 16:50:38 L wasn't the right plaef Apr 09 16:50:42 place rather Apr 09 16:50:52 then stop moaning :P Apr 09 16:50:52 besides, I didn't get server-class hardware until two weeks ago Apr 09 16:51:08 Oh, I'm not moaning... in fact I just got my tickets for Brussels Apr 09 16:51:14 I wish the hotel was cheaper though... Apr 09 16:51:14 cool Apr 09 16:51:23 130euro a night, and 150euro on weekends Apr 09 16:51:27 there is a discount pointed out onm the wiki Apr 09 16:51:32 that's WITH the discount Apr 09 16:51:36 wow Apr 09 16:51:48 ogra : Yeah, makes me wish I had an apartment in Brussels Apr 09 16:52:06 you should move to a place with a decent currency ;) Apr 09 16:52:08 for 1600 euros.. I could rent a REALLY nice place for a month Apr 09 16:52:34 then it hurts less Apr 09 16:52:41 has nothing to do with the currency .. it's the venue, taking advantage that there are three conferences in Brussels all at once, and one of them is UN related Apr 09 16:52:50 ah Apr 09 16:52:58 ogra : Canonical pays for you, Smooth-Stone pays for me :) I' Apr 09 16:53:05 I'm just ... frugal ... Apr 09 16:53:08 the good thing is that the hotel is in the middle of nowhere Apr 09 16:53:21 and anything I don't have to pay for hotel, could be used for more entertaining purposes Apr 09 16:53:27 so you cant spend money running around in the city Apr 09 16:53:31 BAH, humbug Apr 09 16:53:43 like the one they picked for Dallas. Apr 09 16:54:01 that was relatively close to the city Apr 09 16:54:13 thats what happens when you go to ubuntu stuff :P Apr 09 16:54:17 we had worse hotels Apr 09 16:54:43 the paris one was 40km away ... next to the airport Apr 09 16:55:00 and a beer at the hotel bar was 9euro Apr 09 16:57:28 in some conf they made our distro build the stand :D Apr 09 17:03:34 ogra: what are the bootargs you use on the live image for omap? Apr 09 17:04:32 ndechesne: quiet splash vram=12M omapfb.mode=dvi:1280x720MR-16@60 + seed + -- boot=casper Apr 09 17:04:35 for live images Apr 09 17:04:40 quiet splash vram=12M omapfb.mode=dvi:1280x720MR-16@60 file=/cdrom/preseed/ubuntu.seed -- boot=casper Apr 09 17:04:50 lool: you were faster ;-) Apr 09 17:04:52 schnipschnap Apr 09 17:04:53 ubuntu.seed is wrong btw Apr 09 17:04:56 lool, ogra: thanks Apr 09 17:04:58 fail Apr 09 17:05:07 tomorrows image will have the right seed name Apr 09 17:05:12 ndechesne: Do you have a reference of SYS_BOOT information passing between ROM and x-loader? Apr 09 17:05:20 ndechesne: Is this in some public doc somewhere? Apr 09 17:05:28 ogra: basically, what bootargs should I use to try to boot the image on the Zoom2/3 Apr 09 17:05:52 ndechesne, i havent succeeded yet to boot it on the zoom Apr 09 17:05:55 lool: i think it should be in the OMAP3 TRM Apr 09 17:06:13 i suspect i use the wrong load adresses for initrd and uimage Apr 09 17:06:14 ogra: up to where are you going? Apr 09 17:06:21 ndechesne: ok thanks Apr 09 17:06:30 i tried the defaults that were in the nand Apr 09 17:06:55 0x8c100000 for uImage and a guessed 0x82000000 for uInitrd Apr 09 17:07:17 but the kernel doesnt uncompress Apr 09 17:11:50 i suspect i'm overwriting something somewhere Apr 09 17:12:23 omap4 kernel, omap4 hardware not working with the current build Apr 09 17:12:27 upstart is failing Apr 09 17:12:28 ogra: where did you get these addresses? it's been a while since i haven't used my zoom Apr 09 17:12:48 martyn: omap4? which kernel are you using? Apr 09 17:12:49 I'm going to recreate the rootfs, and try again Apr 09 17:13:09 ndechesne: Same as the standard ubuntu kernel, but patched with omap4 support. It worked last week, so something changed Apr 09 17:13:20 I can single mode, so it's not the kernel Apr 09 17:13:22 ndechesne, the 0x81c00000 was the default load address in uboot in nand Apr 09 17:13:33 the 0x8200000 is just a guess Apr 09 17:14:19 ogra: but you need to put the uImage and initrd in DDR, not NAND, right? Apr 09 17:14:46 martyn: where did you get the OMAP4 patches? I am asking because there are many places ;-) Apr 09 17:15:55 martyn: I have 10.04 latest + UNE running on OMAP4. I built root fs with rootstock (e.g. it's not a live image), and I am based of TI OMAP4 kernel: http://dev.omapzoom.org/?p=integration/kernel-omap4.git Apr 09 17:16:04 ndechesne, i fatload them Apr 09 17:16:42 Yeah, I've been skipping some of the kernel-omap4.git patches, because they start forking from mainline too far Apr 09 17:16:56 and that means I can't get other patches to work, like the OpenVZ patch set Apr 09 17:17:04 ogra: ok, so dst address need to be in DDR range, and 0x82000000 is not Apr 09 17:17:47 well, even if i omit uInitrd and only load to 81c00000 it doesnt uncompress Apr 09 17:19:04 ogra: i usually put the uImage at 80300000 or 80200000 Apr 09 17:19:58 http://paste.ubuntu.com/411706/ Apr 09 17:20:09 * ogra_cmpc tries 80200000 Apr 09 17:21:16 ogra: oops I forgot you are on Zoom3. Zoom3 has a different CPU than beagle (3630 vs 3430). I am not sure your current kernel has full support for 3630. can you try on zoom2? Apr 09 17:21:58 yeah, need to replace the CPU though Apr 09 17:26:29 ah, only 256M Apr 09 17:26:35 *sniff* Apr 09 17:27:01 ogra: just tried on my Zoom2, and the kernel boots at @80200000 Apr 09 17:27:09 ogra: yes 256Mb ;-( Apr 09 17:28:51 nope Apr 09 17:29:05 fatload mmc 0:1 0x80200000 /casper/uImage Apr 09 17:29:08 bootm 0x80200000 Apr 09 17:29:09 ... Apr 09 17:29:16 Starting kernel ... Apr 09 17:29:18 nothing ... Apr 09 17:29:19 ogra: on zoom2 or zoom3 Apr 09 17:29:45 zoom2 Apr 09 17:30:01 on my side, kernel boots until it tries to mount the root fs Apr 09 17:30:01 i just changed the SOM back to the one that was in the device Apr 09 17:30:31 hum, it could be your uboot/mlo Apr 09 17:30:38 NAND OMAP34XX ZOOM2 # Apr 09 17:30:45 in any case 34xx Apr 09 17:30:59 if i can trust the installed uboot Apr 09 17:31:17 U-Boot 1.1.4-dirty (Mar 27 2009 - 11:27:11) Apr 09 17:31:34 U-Boot 1.1.4 (Jun 5 2009 - 15:00:50) Apr 09 17:31:35 x-loader doesnt generate any output Apr 09 17:31:44 hmm Apr 09 17:32:36 so I just need to fatload uImage, and initrd? and use the bootargs you gave me, right? Apr 09 17:32:44 right Apr 09 17:32:57 and indeed buse both load addresses with bootm Apr 09 17:33:07 so it switches to the initrd Apr 09 17:33:43 (there is a boot.scr in the image toplevel dir) Apr 09 17:33:54 what do you mean? can you give your commands? Apr 09 17:34:14 fatload mmc 0:1 0x80000000 /casper/uImage Apr 09 17:34:14 fatload mmc 0:1 0x81600000 /casper/uInitrd Apr 09 17:34:14 setenv bootargs quiet splash vram=12M omapfb.mode=dvi:1280x720MR-16@60 file=/cdrom/preseed/ubuntu.seed -- boot=casper Apr 09 17:34:14 bootm 0x80000000 0x81600000 Apr 09 17:34:25 thats the default boot.scr on the image Apr 09 17:34:30 ogra_cmpc, how does the kernel where to find the root image? Or is the initrd all that's needed? I mean how can the system know that it needs to access the /casper/filesystem.squashfs on /dev/mmc0p1... Apr 09 17:35:03 the initrd scans all devices it finds for that file Apr 09 17:35:57 if you set boot=casper on the cmdline the casper script is executed... casper is the tool that sets up the live environment and mounts the squashfs in an aufs mount Apr 09 17:36:13 ogra_cmpc, Would it be possible for you to post a console output of a working target? Apr 09 17:36:41 not atm, the zoom doesnt boot and i dont have any other target around Apr 09 17:36:50 :D sure, np Apr 09 17:37:03 nosse1, but your paste clearly shows you end up in the initrd Apr 09 17:37:08 or :( is perhaps more correct Apr 09 17:37:23 Loading, please wait... Apr 09 17:37:35 thats the first line the initramfs writes Apr 09 17:37:46 try booting with break=top Apr 09 17:37:57 and see if you end up in a busybox shell Apr 09 17:38:26 Yes, clearly initrd, however I doubt is has completed and ready to pass over to rootfs Apr 09 17:38:30 I'll try, thanksw Apr 09 17:38:32 nosse1, also how long did you wait for a console Apr 09 17:38:58 it might be very slow on 256M Apr 09 17:39:32 (its relatively decent on beagle though) Apr 09 17:40:00 AARRRRGGGGHHHH! Apr 09 17:40:16 omap4 didn't boot because one of the NAND chips has a bit error on it Apr 09 17:40:24 and that error is in the first block. Fuck Apr 09 17:41:54 ndechesne, you really need to update to a decent uboot that 1.4.x series is just annoying (hush shell support rules) Apr 09 17:42:03 ogra: this is weird, if i set bootargs, then my kernel does not boot. if I run bootm without setting the bootargs it does boot.... Apr 09 17:42:26 try dropping the video stuff Apr 09 17:42:34 thats clearly for beagle Apr 09 17:42:39 ogra: i know... the problem when you don't stick to mainline, is that the more you wait the more painful it is to catch up ;-) Apr 09 17:42:48 yeah Apr 09 17:43:02 I didn't get any output from kernel until I added console= (and ogra later told to add serialtty=) Apr 09 17:43:25 serialtty is really only for getting a login prompt Apr 09 17:44:21 https://edge.launchpad.net/ubuntu/+source/linux-ti-omap/2.6.33-500.5/+build/1684225 Apr 09 17:44:28 sweet ... kernel finished Apr 09 17:45:13 Argh. The break=top had no effect Apr 09 17:45:35 "Loading please wait..." is the last lifesign from the board Apr 09 17:45:59 you cond get the ramzswap stuff anymore > Apr 09 17:46:01 ? Apr 09 17:46:08 *dont Apr 09 17:46:15 Dont Apr 09 17:46:23 sigh, that cmpc knbd is so small Apr 09 17:46:27 *kbd Apr 09 17:48:20 BTW: How can I force a module to load during boot? I thought editing /etc/modules would, but looking in the initrd, there's no conf/modules file present Apr 09 17:49:10 there is a special /etc/initramfs-tools/modules you can use Apr 09 17:49:20 but you need to rebuild the initrd Apr 09 17:49:25 sure, thanks Apr 09 17:51:41 What is the real difference between update-initramfs's -c (create) and -u (update)? Apr 09 17:53:28 u updates and creates a backup of the old one Apr 09 17:54:25 ogra: still something weird on my board. if I don't set the bootargs and load uimage and uinitrd, i can get the initramfs prompt Apr 09 17:55:59 what bootargs do you use now ? Apr 09 17:56:48 nevermind... it's too late ;-) i forgot the console stuff. Apr 09 17:56:59 ndechesne, the only arg thats actually necessary is boot=casper Apr 09 17:57:36 you shouldnt need console if you have any external display you can drive Apr 09 17:57:39 with setenv bootargs 'console=ttyS3,115200n8 mem=256M splash file=/cdrom/preseed/ubuntu.seed -- boot=casper', kernel boots, but omapfb fails, so there is no display. Apr 09 17:57:50 yeah Apr 09 17:58:08 try adding serialtty=ttyS3 Apr 09 17:58:19 what does it bring? Apr 09 17:58:29 should spawn a getty on ttyS3 Apr 09 17:58:46 so you get a serial login Apr 09 17:58:54 ok Apr 09 17:59:54 with the normal bootargs, on zoom2, kernel crashes with some usb problems.. but I am using the old .img from 2 days ago.... Apr 09 18:00:13 oh Apr 09 18:00:19 yup ... ;-) Apr 09 18:00:34 yeah, use the img from my people.canonical.com page Apr 09 18:00:44 that has a kernel with fixed USb Apr 09 18:01:37 i need to go now... i will try the new image on monday ... but I think that the new image will not resolve the omapfb issue. i will look into that on monday. Apr 09 18:02:06 yeah, unliklely that omapfb works on anything but beagle i guess Apr 09 18:02:18 ndechesne, you say Zoom2, is that the beagle you're talking about? Apr 09 18:02:31 nosse1, nope thats the zoom Apr 09 18:02:43 http://omapzoom.org/ Apr 09 18:03:02 What CPU? Because this kit (AM3517-EVM) is also a Zoom kit (IMHO) Apr 09 18:03:22 nosse1: zoom2 has 3430, zoom3 has 3630. Apr 09 18:03:29 ogra: is that the one: http://people.canonical.com/~ogra/lucid-netbook-armel+omap.img Apr 09 18:03:58 right Apr 09 18:04:13 ok. i will start download now, and leave, bye. Apr 09 18:04:17 you should have usb Apr 09 18:04:22 have a nice weekend Apr 09 18:04:47 http://www.logicpd.com/products/development-kits/zoom-am3517-evm-development-kit Apr 09 18:04:52 * ogra_cmpc also calls it beer'o clock now Apr 09 18:05:13 Yeah! for good product (re)naming.... Apr 09 18:13:20 * ogra_cmpc slaps forehead Apr 09 18:13:23 oh my Apr 09 18:13:46 console=/dev/ttyS3,115200n8 Apr 09 18:13:51 indeed that cant boot Apr 09 18:14:07 * ogra_cmpc wonders how that got in the default env Apr 09 18:17:54 heh Apr 09 18:18:58 sweet, it seems to boot Apr 09 18:20:10 hmm, no, boot ends at squashfs Apr 09 18:20:49 oh, its just slow Apr 09 18:21:26 * ogra_cmpc twiddles thumbs Apr 09 18:21:53 wohoo Apr 09 18:22:20 nosse1, so on the zoom2 i get a boot but it takes about 2 min Apr 09 18:22:46 ubuntu@ubuntu:~$ cat /proc/cpuinfo |grep Hard Apr 09 18:22:46 Hardware : OMAP Zoom2 board Apr 09 18:23:19 setenv bootargs console=ttyS3,115200n8 serialtty=ttyS3 file=/cdrom/preseed/ubuntu-netbook.seed -- boot=casper Apr 09 18:23:21 mmcinit Apr 09 18:23:28 fatload mmc 0:1 0x80200000 /casper/uImage Apr 09 18:23:36 fatload mmc 0:1 0x81000000 /casper/uInitrd Apr 09 18:23:42 bootm 0x80200000 0x81000000 Apr 09 18:23:50 that gets it going for me Apr 09 18:25:12 * ogra_cmpc goes to try 3630 now Apr 09 18:25:23 It gets going for me as well, except it stops somewhere down the line Apr 09 18:25:55 well, its takes quite long Apr 09 18:26:01 2-3 min Apr 09 18:26:02 On my other test (trying to run roostocked lucid), it stops in the initrd on "+ wait-for-root /dev/mmcblk0p1 30" Apr 09 18:26:30 BTW: Why do you explicitly specify mem=xxM ? Apr 09 18:26:42 i dont ? Apr 09 18:26:50 well you don't, but I've seen it around here Apr 09 18:27:03 it was a general q Apr 09 18:27:19 you normally dont need to Apr 09 18:27:37 the kernel handles that fine Apr 09 18:30:14 * nosse1 is giving it yet another go. Stopwatch on Apr 09 18:31:38 How many minutes until declared dead? Apr 09 18:33:05 max 5 i'd say Apr 09 18:33:37 amitk, any chance we can have 3630 together with 34xx in our kernel ? Apr 09 18:34:23 Do you guys use any JTAG tools in case of deep crash, or do you rely solely on SW mechanisms? In case which tools do you use? Apr 09 18:35:14 * ogra_cmpc didnt have to debug such deep crashes yet Apr 09 18:35:30 and all targets we support are usually unbrickable Apr 09 18:35:54 i.e. they support getting the bootloader from SD Apr 09 18:36:22 nosse1: i personally recommend the flyswatter and openocd, but i'm biased Apr 09 18:36:35 * XorA recomands flyswatter as well Apr 09 18:36:40 Ah. Point is we are bringing forth new HW based on the OMAP, so for the custom driver development we need JTAG for observability. I was curious what other linux devlp. are using Apr 09 18:36:45 XorA: hehe thanks Apr 09 18:37:31 * ogra_cmpc is a glue guy ... my work usually starts above kernel and with already existing bootloader sources Apr 09 18:37:35 nosse1: openocd is not as robust as something like a BDI or lauterbach, but it generally works well enough to debug most issues with driver development Apr 09 18:38:50 I had a demo one of lauterbach with extremely smoooth kernel & userspace integration, however my boss wont agree on the pricetag :D Apr 09 18:38:57 s/one/once/ Apr 09 18:41:03 ogra_cmpc: yes, you can have 3630 support in the same kernel Apr 09 18:43:24 13 mins passed. I think it's dead. Apr 09 18:43:40 * nosse1 would be happy if he had a JTAG tool to see if target is really dead Apr 09 18:44:04 nosse1: I've used lauterbach in my previous job, I've just acquired the tincan tools jtag, haven't tried it yet Apr 09 18:45:23 A local company is making a new tool named ZY1000: http://www.zylin.com/zy1000.html Apr 09 18:46:27 ...oh forgot - we've switched to A8 - which isn't listed yet Apr 09 18:49:16 Since the boot hangs on "+ wait-for-root /dev/mmcblk0p1 30", how can I modify Ubuntu for NFS root? Apr 09 18:49:53 Edit /etc/fstab, put up the NIC driver in /etc/initramfs/modules. More? Apr 09 18:52:30 * nosse1 will try https://help.ubuntu.com/community/DisklessUbuntuHowto Apr 09 18:53:41 nosse1: openocd is free, and the flyswatter retails for $49.00 Apr 09 18:54:11 prpplague, yeah, I saw that. Cheap! Apr 09 18:54:29 nosse1: i prefer inexpensive! Apr 09 18:55:59 prpplague, hehe we don't have that distinction in my language... Anyway they ship to Europe I hope Apr 09 18:56:20 nosse1: we do Apr 09 18:56:23 heh, defineately not cheap Apr 09 18:56:31 nosse1: the USPS service is the cheapest shipping Apr 09 18:56:45 _we_? Apr 09 18:57:21 amitk, just tested your 500.4 link from this morning, the ehci port looks good on my C4 board. Found all the random assorment of usb based devices... Apr 09 18:58:22 nosse1: i said i was biased Apr 09 18:58:27 * prpplague designed the flyswatter Apr 09 18:58:38 lol Apr 09 19:00:33 prplague, I'd like to talk more about the flyswatter. However I haven't got the time now. Can I PM you later on this? Apr 09 19:01:15 nosse1: you can feel free to message me over in #edev to discuss it at any time Apr 09 19:01:27 nosse1: or pm me Apr 09 19:01:31 Thanks Apr 09 19:13:34 YYYEEESSS!!!!! I finally got Ubuntu up and running on target! Apr 09 19:13:40 * nosse1 VERY HAPPY! Apr 09 19:13:57 what was it ? Apr 09 19:14:03 just slowness ? Apr 09 19:14:43 it's the kernel. I have compiled a new kernel using a newer one from TI Apr 09 19:14:51 ah Apr 09 19:15:22 And I'm running NFS root. The MMC is probably not working on the vanilla omap target Apr 09 19:15:44 works on the beagle and zoom Apr 09 19:16:01 @ogra: how is the Zoom coming along? Apr 09 19:16:43 cpearson, well, the 34xx version boots slower than the beagle, the 36xx one doewsnt work with our kernel Apr 09 19:17:06 same issue as with the XM? Apr 09 19:17:27 it just doesnt have 36xx support built in Apr 09 19:17:37 but amitk just said its possible Apr 09 19:17:51 so if we can we will enable it Apr 09 19:19:04 beyond that, our current image is indeed targeted for beagle so the framebuffer etc have issues Apr 09 19:31:15 rcn-ee: \o/ Apr 09 19:31:49 ogra_cmpc: is 36xx support a priority over OTG/PM/ etc.? Apr 09 19:32:07 amitk, is there anything else on your check list? I'm downloading ogra's latest *img file.. Apr 09 19:32:18 making up a sd card for the xm to see how bad it is. ;) Apr 09 19:32:37 rcn-ee: XM is on my wishlist next Apr 09 19:32:54 but I need to do some family time (it being friday evening and all that) Apr 09 19:34:54 of course family first ;) My goal is to get this XM booted and benched against it's older boards this weekend (inbetween beer and golf).. Apr 09 19:35:21 beer, great idea. Apr 09 19:36:23 and someone broke gcc trunk... i think i need 2-3 more beagles for faster bisecting.. Apr 09 19:37:36 isn't trunk expected to be broken a bit sometimes, now after branching 4.5? Apr 09 19:38:22 yeah.. it's perfectly fine at this stage.. ;) Apr 09 19:38:49 OK. I got lucid running, but only with custom kernel (based on non-ubuntu source). Apr 09 19:39:11 The last kernel ti-omap -5 does not work on AM3517, see http://pastebin.com/RdiMijNL Apr 09 19:39:13 hey nosse1, which kernel repo did you use again? Apr 09 19:39:35 For the one that works, I'm using TI's git repo Apr 09 19:40:09 So, I'm ready to conclude that the official kernel wont work for my target. Apr 09 19:40:18 actually that's not bad... nosse1 does the console ever come up with 500.4? Apr 09 19:40:27 But it's not tragic, because we will have custom HW (hence kernel) anyways Apr 09 19:41:43 The cheering from me was for running Ubuntu userspace! Apr 09 19:42:29 nosse1, is that their psb/bsp repo? I'm going to steal/browow some xm stuff Apr 09 19:44:51 rcn-ee: It's this repo: git://arago-project.org/git/people/sriram/ti-psp-omap.git I know TI has two repos: This one and the one on the official kernel server Apr 09 19:45:23 According to the other developers here the former is the most updated one, while the last is for what is pushed upstream Apr 09 19:45:51 Thanks nosse1, wasn't 100% where it was, I normally pull from tmlind's tree and after 2.6.31 i kinda was off in the woods doing my own thing.. ;) Apr 09 19:46:44 I haven't a full understanding of the difference between those two repos. Apr 09 19:47:05 We need to base our HW branch from one of them, but I'm not sure which yet Apr 09 19:47:33 talking with my buddies at TI it's their new, more upstream kernel policy then years past, 'plan'.. Apr 09 19:49:17 I spent some considerable time to ubuntuize the kernel config for the ti-kernel Apr 09 19:50:34 It's kind of a challlenge, in years past you've allways had one specific config for one specific board, but now for distrutors, you need one config for all.. Apr 09 19:51:34 Now we are ready to evaluate _if_ we are going to run Ubuntu at all on the device... Apr 09 19:51:57 ogra: ping Apr 09 19:52:14 i can give you one hint... it's faster then debian.. ;) Apr 09 19:52:38 Thanks, that answers one ;) Apr 09 19:53:42 I need to get the SVX-stuff and Qt on top (either fb or X11) to eval the end user GUI Apr 09 19:54:59 the sgx modules are pretty easy to build, i have a couple scripts in my bzr repo... Haven't messed around with the QT-SGX setup thou... Apr 09 19:55:49 bzr repo? Apr 09 19:56:27 sorry: https://launchpad.net/~beagleboard-kernel in the 2.6-stable and 2.6-dev tree's theirs a build_sgx_modules.sh script... Apr 09 19:58:52 Can one cross-compile apps for Ubuntu target? Apr 09 19:59:32 Some of our developers are worried about having to compile Qt on target for the target, while we have a rather large build farm Apr 09 20:00:01 Yeah, it's not always 100%, but it's fast for a quick test, i usually build on either a qemu based chroot or native chroot... Apr 09 20:00:08 nosse1: qt is going to be very slow if you compile it inside qemu Apr 09 20:00:19 nosse1: cool to see that you finally made to boot ubuntu :-) Apr 09 20:00:26 Oh yes! Tried it! Apr 09 20:00:51 rsalveti, hi yes. I got it to boot only an half hour ago! Finally! Apr 09 20:01:45 nice :-) Apr 09 20:02:22 nosse1: you could probably use your build farm inside qemu, but you have to setup the cross compiler correctly Apr 09 20:02:49 and depends on what system you're currently using Apr 09 20:03:18 Yes. I'm running qemu/chroot to update and prepare packages from target, and that work fine (however slowly) Apr 09 20:03:24 at least with sbox I already saw it working with icecream Apr 09 20:04:26 c++ sucks even when you build it at your native invironment Apr 09 20:04:41 *environment Apr 09 20:04:54 Should I use serialtty=ttyS2 on installed Ubuntu as well? Apr 09 20:05:24 (I solved this by adding my own /etc/init/ttyS2.conf instead) Apr 09 20:14:54 Try telling the app developers that C++ sucks.... Apr 09 20:17:22 Interesting: https://code.launchpad.net/~beagleboard-kernel/+junk/2.6-stable, clicking on a revision like this: http://bazaar.launchpad.net/~beagleboard-kernel/+junk/2.6-stable/revision/64 then I get internal server error Apr 09 20:20:18 yeah noticed that too.. something wrong with launchpad.. click reload works for me... Apr 09 20:26:03 rcn-ee, how do you experience Ubuntu against say Arago/OE or any other cross compiled into-one-large-image distro? Apr 09 20:27:33 size is obvious, but we've got nearly 1G available for entire system, so that's handled Apr 09 20:27:42 well, from my experience (pre lucid) Angstrom has always been the fastest and most polished, mostly due to many arm developers working on it for many years... Apr 09 20:28:21 other then the 'feel' i haven't run any numbers for load comparisons yet... Apr 09 20:32:02 one interesting side note, i do have a touchbook running both Angstrom and ubuntu Karmic... Although Karmic is slower, more things do work out of the box, vs some of hte applications in the Angstrom image.. Apr 09 20:33:25 Yes, that another point. I expect customer upgrade/maintenance/security fixed to be easy with Ubuntu. I have no idea what that would imply on Ångström Apr 09 20:34:53 They use a tool called ipkg very similar to dpkg, but i haven't figured it out 100%.. Long term maintance is ubuntu/debian's strong point... Apr 09 20:35:40 Yeah. If we can stick with vanilla Ubuntu arm, and then our own custom kernel + our app, then I'd be /very/ happy Apr 09 20:38:21 Theres one product requirement though. Boottime from off until GUI is shown must be minimized, so perhaps some customization to the boot process must be added as well Apr 09 20:39:34 there is the 10second wait in most u-boot that's easy to remove... Apr 09 20:43:42 Timing it now, it takes 21 secs from u-boot bootm to login is shown (i.e. kernel and initrd loading is not included). This is with NFS root and all so I'm very optimistic about shaving off seconds. Apr 09 20:44:48 console login or gui? Apr 09 20:45:03 console. I have no gui as of now Apr 09 20:45:38 okay, i had a user report that "console-setup-mini" shaved some more seconds in the lucid alpha releases... Apr 09 20:46:41 rcn-ee, perhaps you know what is needed to get the display up and running (I have the AM3517-EVM)? Apr 09 20:47:53 i'm comparing the ti-psp-omap repo with ubuntu's. to figure out xm stuff ;) what git checkout did it work in ti-psp's? Apr 09 20:48:36 Uhm. /me git newbie: How to check? Apr 09 20:49:25 ti does things like OMAPSP_03.00.00.04/etc.. which will get us in the ball park... Apr 09 20:49:50 No, I've pulled their working git repo for the kernel Apr 09 20:51:52 well they have about 12-15 more commits vs ubuntu's... your using the LCD panel? Apr 09 20:53:49 I'm about to... Apr 09 20:54:47 I'm going to try to merge them in.. Apr 09 20:55:40 I can give you my head git revision if I just knew how to get it Apr 09 20:56:26 git log (top one sha) is the easist that i know off Apr 09 20:57:10 This one 7b8926aa626 Apr 09 21:03:13 okay it's going to take quite a bit more cherry picks, but i think i got the lcd, it'll take about 5mins to finsih building.. Apr 09 21:04:10 yeah, it's going to take more... it doesn't like the board-am3517evm at all... it's just too new with very little upstream... Apr 09 21:07:36 Yeah, I think I have logged over 50 hrs on this thing now. But part of that is to get oneself up to speed on Ubuntu booting internals Apr 09 21:07:53 So the positive result this evening is most welcome :D Apr 09 21:08:50 it's very possible and doable, both tree's are based off 2.6.33, but ti has about 100 new patches.... Apr 09 21:09:17 i really hope they merge some of that inthe next cycle... Apr 09 21:22:43 * nosse1 is trying xserver-xorg-video-omap3 Apr 09 21:29:11 Adam want long in Eden: [ 5577.017395] BUG: soft lockup - CPU#0 stuck for 61s! [dpkg:681] Apr 09 21:29:18 *wasn't Apr 09 21:54:03 * NCommander isn't doing good. Apr 09 22:08:17 does dave martin frequent this channel? Apr 09 22:08:45 (I need someone to look at some gutsy v7 asm) Apr 09 22:53:11 I see I get a lot of unresolved packages when I try to get "ubuntu-netbook". Apr 09 22:53:34 What's a good target for the standard gnome desktop? **** ENDING LOGGING AT Sat Apr 10 02:59:57 2010