=== DanaG1 is now known as DanaG [03:43] * Add a new 'fixrtc' script that tries to set the system clock forward based on the last mount time of the root disk; without this kludge, systems without a working RTC will end up in a perpetual reboot loop. Thanks to Dave Martin for the patch. LP: #563618. [03:43] nice [04:21] hmm, now how do I make the beagle use the flashkernel kernel? [04:24] And what bootargs does it want? [04:40] DanaG: flash-kernel *should* write the kernel to the SD card and set the bootargs. Which bit isn't working? [04:41] Hmm, does it write to SD, or to NAND, or both? [04:42] * persia checks, but thought it wrote to SD [04:43] Can't find /boot/vmlinuz---help and /boot/initrd.img---help [04:43] HAR. [04:44] Seems to be NAND only, actually. [04:45] Aha! That's a different issue. [04:57] I looked in flash kernel, and don't see where it sets boot args. [04:58] Neither do I, really. [04:59] TO me it looks like it runs mkimage, adn then copies the result. === Baybal__ is now known as Baybal === hrw|gone is now known as hrw [08:02] morning [08:41] lool, https://bugs.edge.launchpad.net/ubuntu/+source/rootstock/+bug/570588/comments/6 do you have any idea about that one ? [08:41] Launchpad bug 570588 in rootstock (Ubuntu) (and 1 other project) "/usr/bin/rootstock: line 195: 13642 Segmentation fault qemu-system-arm $QEMUOPTS -append "${APPEND}" > $QEMUFIFO 2>&1 (affects: 1) (heat: 6)" [Undecided,New] [08:43] * ogra hasnt seen anything like that [09:29] ogra: Hey [09:29] ho [09:30] ogra: Just FYI I tested the lucid omap image on my beagleboard and it didn't boot [09:30] I think the boot script uses the source command which I dont have support for in the default u-boot [09:30] ugh, thats a C4 ? [09:30] yes [09:30] I dont think I ever replaced the u-boot in flash [09:31] https://wiki.ubuntu.com/ARM/BeagleNetbookInstall should work then [09:31] i'm sure all C4 u-boots have hush shell support [09:31] I actually have a box of c4s here [09:31] * ogra just looks at http://h10010.www1.hp.com/wwpc/us/en/sm/WF06a/12454-12454-321959-338927-3640405-4063703.html [09:32] HP does a very good job of not giving *any* info about the marvell CPU used in these devices [09:33] Marvel armv7a lacks neon [09:33] lool, can you try the stuff under "Preparation work" wrt uboot on the wikipage [09:33] hrw, yes, but we have armada/dove support while we dont have kirkwood support in ubuntu [09:33] lool: which uboot ver you have on c4? [09:34] that device might make a nice dev platform if it would be dove [09:36] this is not a kirkwood I think. [09:36] never herd of kirkwood with gfx inside [09:36] well, i would expect the same regarding the 1.2GHz [09:36] ogra: sorry, it's a C3 not a C4 [09:36] I have C4 here with which we will try [09:36] openrd-base/client uses XGI chipset for video [09:36] but there is no concrete doc that tells anything about the CPU at all [09:37] armada 100/500/1000 fits [09:38] lool, C4 should work out of the box ... though note that flash-kernel rewrites the uboot config and there is no "restore" command (i didnt have teh time to rewrite it to use teh existing u-boot vars, i'll do that for 10.0+), better save the printenv output if you want to restore [09:39] * ogra will add fw_backup_env and fw_restore_env commands to uboot-envtools on 10.10 [09:40] ogra: newegg has them cheaper: http://www.biz.newegg.com/Product/Product.aspx?Item=N82E16859105659 [09:41] hrw: iirc that hp is essentially openrd-client [09:41] suihkulokki: but openrd-client has separate videoram iirc [09:42] hrw: like that hp has [09:42] heh, "you save $20" ... "shipping costs $10" [09:42] funny [09:43] http://www.linuxfordevices.com/c/a/News/HP-t5745-and-HP-t5325/ says kirkwood [09:44] http://my.opera.com/bhtooefr/blog/hp-t5325-thin-client-risc-os-and-maybe-combining-the-two-or-just-running-linu [09:44] includes link to dmesg [09:44] 1.2 GHz Marvell 88F6281 Kirkwood SoC aha [09:45] so its a sheeva on steroids [09:45] not ubuntu then :/ [09:46] yep [09:47] ogra and hrw, mouser cancelled our order for beagle board [09:48] who is mouser ? [09:48] because they cannot ship it to china due to some US export law [09:48] bah [09:48] or mouser is a similar courier like digikey [09:48] or distributor [09:48] cant they ship it to europe/us office and then let company take care of it? [09:58] cooloney, if you'd like to I can try to arrange you cheap cm-t35 (http://compulab.co.il/t3530/html/t3530-cm-datasheet.htm [09:59] Mike^: oh, man, that is nice [09:59] err [09:59] note that the only omap we dont support is the cm-t35 [09:59] (in the omap3 series) [10:00] ogra, right :) great thanks :) [10:00] Mike^: as ogra said, we can ramp up ubuntu on cm-t35 [10:00] http://ograblog.wordpress.com/2010/04/23/unleash-the-beagles/#comment-246 [10:01] * ogra just was pointed to the fact a minute ago [10:01] i'll have to ask amit why he didnt enable it, i guess he had reasons for it [10:01] ogra, maybe becase it has display support merged after 2.6.33 [10:01] ah, yeah, that might be [10:02] an i guess it doesnt come with the 256M it supports by default, does it ? [10:03] amitk, heh, speaking of the devil [10:03] amitk, http://ograblog.wordpress.com/2010/04/23/unleash-the-beagles/#comment-246 [10:03] ogra, it has 256M by default [10:03] ah, the spec says 64-256 [10:03] but can be assmebled with 64 and 128 if the customer will ask [10:03] ok [10:04] ogra, possible display is not the issue, because e.g iegp2 has no display either in 2.6.33... [10:04] Mike^: is there any big difference between cm-t35 and beagle board? [10:04] maybe kernel is not a big problem, [10:04] ubuntu needs display, right? [10:05] depends what you want to do with it :) [10:05] we have netinstall and server images for omap to use it headless [10:05] cooloney: t35 has TPS69030 instead of TWL4030 that beagle has, but on the other side it has on-board wifi and Ethernet :) [10:05] ogra: I know about it, there's some bug with the alsa init in the CM_T35 board, making the driver modular doesn't work. So I had to remove it completely. [10:06] amitk, mind to comment ? [10:06] If someone has a fix, I'll take it. Else I'll need some time to fix it [10:06] Mike^: yeah, that is better, i think [10:06] amitk: i think Mike^ is the man [10:06] Mike^: tps69xxx and twl4030 use same drivers [10:08] hwr, right, but tps69030 has less features than twl4030 [10:11] cooloney: I'll talk to our sales guys and send you an e-mail after that [10:11] amitk: do you remember what was the bug with audio? [10:14] * XorA|gone points to the ASoC maintainer sitting on this channel :-) === XorA|gone is now known as XorA [10:15] ;DD [10:17] ogra: sorry, got distracted, machine crashed and more [10:17] ouch [10:17] ogra: So my board doesn't boot of the box with the lucid image [10:17] but a C4 out of digikey does [10:18] ogra: So you're all good [10:18] It's probably specific to C3 [10:18] older u-boot or so [10:18] lool, and for your board, just look at boot.scr and type the commands on the uboot prompt :) [10:18] ogra: Well ISTR it was "source" failing [10:18] right, i'll write the flash-uboot-x-loader wikipage today :) [10:19] so you can upgrade if you want to [10:19] ogra: http://www.angstrom-distribution.org/demo/beagleboard/README.txt is good base for it [10:19] luckily we dont need hush shell support in our setup [10:20] hrw, hmm, for the variant on the prompt at least, i aslo want to do it from a running system [10:20] Mike^: thanks a lot, man. heh [10:21] but first i have to test server and netinst images :) [10:25] lool, btw, did you see my ping from this morning ? [10:25] ogra> lool, https://bugs.edge.launchpad.net/ubuntu/+source/rootstock/+bug/570588/comments/6 do you have any idea about that one ? [10:25] Launchpad bug 570588 in rootstock (Ubuntu) (and 1 other project) "/usr/bin/rootstock: line 195: 13642 Segmentation fault qemu-system-arm $QEMUOPTS -append "${APPEND}" > $QEMUFIFO 2>&1 (affects: 1) (heat: 6)" [Undecided,New] [10:25] i have never seen something like that [10:33] ogra: that rings a bell [10:33] looks very scary [10:33] ogra: what's triggering this? [10:33] which command? [10:34] ogra: can you remind me rootstock page? I will try [10:34] seems its the start of the VM [10:34] https://wiki.ubuntu.com/ARM/RootfsFromScratch [10:35] note that he uses a script with bashisms (i have an SRU already for that, didnt notice when i merged the patch) but thats shouldnt affect anything at that point [10:35] (it causes the "...initrd.img-*': No such file or directory") [10:36] time to install lucid in VM [10:37] time to install lucid on your beagles ! [10:38] * ogra still doesnt see many omap related bugs ... i would expect more [10:38] ogra: my C3 refuses to see usb devices [10:38] install to SD :) [10:38] the netinstall offers that [10:38] https://wiki.ubuntu.com/ARM/BeagleNetInstall [10:39] heh, first requirement on that page is a usb device [10:39] * XorA offers ogra more coffee [10:39] *slurp* [10:40] ok, you need a NIC ... indeed [10:40] * ogra slaps forehead [10:40] and it has "attach keyboard" [10:41] well, you can change the comdline in boot.scr and use serial [10:41] but no NIC is indeed a showstopper for a netinstall [10:41] I have 3 dm9601 usb network adapters [10:41] someone hurry up and patch the kernel with zippy support :-) [10:42] send beer to amitk :) [10:42] ogra: and zippy2 [10:42] send more beer to amitk :) [10:42] Mike^: we had some problems booting beagle with CONFIG_OMAP_MUX. But CM_T32 depends on it. So we had to unfortunately disable the board. I hope to re-add that support sometime... [10:43] although I can do what I did for zoom2 and boot lucid off the Angstrom kernel [10:43] sure, you just lose all upgradeability [10:44] but gain a lot of features [10:45] might be but you might be screwed at some point using dist-upgrade [10:46] depends how you installed [10:46] well with no working USB Ive only got one choice to install and that rootstock with minimal image [10:46] lunch [10:46] JaMa: you here too? [10:47] * hrw checks again... this is not #oe channel? [10:47] XorA, yeah, that wont set up the kernel or initramfs [10:47] initramfs seems to be optional [10:48] its not [10:48] ogra: I dont think the bashisms are the issue though :-) [10:48] make sure to never install encryptfs relared stuff :) [10:48] lool, no, i just wanted to mention it [10:48] Ok [10:49] I'm afraid my qemu time shrunk considerably as of late [10:49] since it spills an additional error in the log [10:49] lool, forever or just because of release ? [10:49] i thought its among your new responsibilities to care for virtualization [10:50] XorA: There is a boot bug due to audio init here. And the driver can't be modular. The real fix is to move the omap_ctrl_* calls to the clock framework, IMO. https://bugs.edge.launchpad.net/ubuntu/+source/linux-ti-omap/+bug/556482 [10:50] Launchpad bug 556482 in linux-ti-omap (Ubuntu) "kernel crash when booting on AM3517-EVM (affects: 1)" [Wishlist,Fix released] [10:50] x86(-64) installer for lucid is nice [10:50] amitk: prod lrg about that [10:50] hrw, its the same as for ARM ... [10:51] amitk: he is ASoC maintainer and inventor [10:51] ogra: so far I did not had occasion to do anything with arm one [10:51] ah [10:51] XorA: aah, right. I think I met lrg at Plumbers Conf last year [10:52] ah those good times when ASoC meant WM8xxx/WM9xxx chips used in zaurus devices [10:52] amitk: he was there [10:55] ogra: i've tried netinstall yesterday and it crashed telling something like "no more free space" during packages install [10:55] Mike^, what task did you select ? and how big was your disk ? [10:56] ogra, it was before task selection, as far as I understand it was during debootstrap [10:56] (for a netbook task you should use 4G or more, minimal should have around 500MB at least) [10:56] oh, weird [10:57] how big was your disk then ? [10:57] a have 80G usb disk :) [10:57] hmm, that should be plenty [10:57] having a bug with the installer logs would be helpful [10:57] amitk: the [ 675.154052] Failed to add route LOUT->Line Out suggests the machine driver needs updating too [10:57] ogra, i'll try :) [10:57] though if your HW isnt supported in the kernel it might be related [10:58] ogra, I've added my hw to the kernel I use for boot. And netbook mix works Ok [10:58] hmm, k [10:59] it might be d-i that needs to know about your HW too [11:00] the archdetect command is used in many places in d-i and it gets its info from the Hardware line of 7proc/cpuinfo [11:00] though "no more free space" is a weird error [11:01] i can imagine it doesnt know what kernel to select if your board isnt in archdetect [11:01] but that would be a different error msg [11:01] ogra, I don't remember exactly, I'm going to retry now and see what's going on there [11:02] * Mike^ running Lucid netinstall [11:02] tty4 has the log btw [11:05] * ogra notices the time and considers some breakfast [11:05] bbl [11:05] lrg: ok, I'm not an audio expert. I'll fix the driver to be modular and then ping you for help. [11:29] persia: ping [11:29] saeed: What's up? [11:30] persia: hi there [11:30] perisa: I tried to install openswan [11:30] but it never finish after doing the "Setting up .." [11:32] saeed: You you ran `apt-get install openswan` and it never took you through the certificate setup? [11:32] it did [11:33] OK. At what point isn't it working for you, precisely? [11:35] the apt-get install hangs after printing "Setting up openswan " [11:36] I'm confused. When I install, I get debconf prompts for the certificate setup after "Setting up openswan". [11:37] And you say you got that, but also that it hangs at that point. Could you paste a log or similar? [11:39] persia: I've killed it. I'll retry and send you the log [11:47] perisa: I'm in deadlock, I can't remove or purge openswan as dpkg is interrupted and I must run "sudo dpkg --configure -a" [11:48] OK. What happens when you run `dpkg --configure -a`? [11:48] but then that will try to resume the openswan installation which hangs again [11:49] saeed: Well, let's see if debugging helps :) Try adding "set -x" near the top of /var/lib/dpkg/info/openswan.postinst [11:49] ogra: The netinstall seems to work now. Probably I've run it last time without the USB disk and MMC card has exploded. So, no bug report for now :) [11:50] oki :) [11:54] persia: apt-get remove worked, I tried also purge then re-install, it hangs before reaching the configuration setup [11:55] CTRL+C doesn't help as well [11:56] So, if you purged, did it go through the certificate wizard this time again? [11:58] no [11:58] But it did last time? [11:58] the openswan install hangs at :/usr/lib/ipsec/rsasigkey 2048 [11:59] OK. What happens if you try to run that directly, rather than as part of the postinst? [11:59] I tried strace /usr/lib/ipsec/rsasigkey 2048 [12:00] and it wait on /dev/random [12:00] when I move the mouse [12:00] I can see that it succeeded to read few bytes [12:00] it seems I have no enough random devices on dove [12:00] Do you have a TPM or RNG on the board? [12:01] no [12:02] we don't have random generation device [12:02] Yeah, other than that you're stuck with disk read timings, keypress timings, mouse movements, and network checksums, which may not be much at all :( [12:03] Sometimes running `find / -name *gz -exec gzcat {} > /dev/null;` or similar can help, as it generates lots of IO. [12:03] Or start a torrent. [12:06] persia: sometimes my system boots with network disabled [12:07] I notices those errors on dmesg: http://paste.ubuntu.com/423915/ [12:07] That's unexpected. [12:08] Isn't that just the app_armour profiles? I think that's normal. [12:08] I'm unsure why that would disable your network though. [13:08] minimal ubuntu image is huge [13:08] compared to ? [13:08] OE ones/ [13:08] ? [13:09] thats because ubuntu isnt optimized for embedded [13:09] I know [13:09] debian isnt much smaller [13:09] and i doubt fedora or opensuse are [13:10] but would be nice to fit in 256MB nand on beagle for example [13:10] you can surely achieve that [13:10] rip out the x86 specific parts, remove all docs etc [13:11] i guess that would get you to around 100M [13:11] drop python/perl if possible [13:11] you shouldnt [13:11] *if possible* [13:11] python is an essential piece of ubuntu, many ubuntu specific apps use python [13:12] (tha majority of ubuntu development is done in python) [13:12] and perl is historically used in many apps [13:18] You should be able to get minimal in 256MB *easy* [13:18] 270MB is so far what I have [13:19] Try --no-install-recommends [13:19] 237MB after cleaning /var/lib/apt/lists/ [13:19] Also, use a compressing filesystem on your NAND (e.g. ubifs) [13:20] let me first check what can be safely removed [13:24] hrw, /usr/share/doc ... /usr/share/man [13:25] and as persia mentioned, avoid all recommends [13:25] /usr/share/{x11,locale,doc,man} [13:25] hrw: Really, it's easier to start over. Do your initial install without recommends. [13:25] sure [13:27] ogra@osiris:~$ sudo du -chs /var/build/lucid-arm-chroot/ [13:27] 245M /var/build/lucid-arm-chroot/ [13:27] hmm [13:27] thats with nothing cleaned, plain debootstrapped chroot [13:27] with or without recommends? [13:28] just plain qe,u-debootstrap [13:28] * persia ended up with *exim4* in a sid chroot recently by not remembering to adjust this. [13:28] *qemu [13:28] Right. By default, it includes recommends. [13:28] debootstrap ? [13:28] Yep. [13:29] At least when I debootstrapped sid a few days ago, I ended up with recommends. I haven't debootstrapped lucid in a few weeks. [13:29] Actually, I might have done it only a week ago, but I didn't check for recommends then. [13:30] hmm, minibase should solve that [13:30] my new laptop is in delivery ;d [13:31] ogra: sid debootstarp is 159M after rm /var/cache/apt/* [13:32] Mike^, that doesnt gain you much ... you will at least need to keep the package lists there [13:32] (and apt-get update will recreate them) [13:33] unless you dont plan to do anything with the system :) [13:33] ncdu is nice tool to check space used [13:33] ogra: it's with the lists, just without *debs [13:33] Mike^: 'apt-get clean' is better way [13:33] yeah [13:33] ogra: in embedded one usually doesn't do 'apt-get something' [13:35] hrw: it was a quick test, and 'rm ' is faster than 'chroot /home/mike/arm/sid apt-get clean' [13:35] yep [13:35] ogra@osiris:/var/build$ sudo du -chs /var/build/lucid-minimal/ [13:35] 138M /var/build/lucid-minimal/ [13:35] lucid wins :) [13:35] using: sudo qemu-debootstrap --arch=armel --variant=minbase lucid lucid-minimal [13:36] no idea if thats bootable though :) [13:36] hrw: the best I had with sid is ~350M with gdm and e17 [13:37] ogra: most probably yes, the question if it's usable :) [13:37] if it boots to a shell it will be :) [13:39] for basic arm devel rootfs I would like to have few extras like evtest, tslib-bin [13:40] well, that probably gets you to 150M [13:40] And the budget is probaly ~400MB if you use ubifs [13:41] less if you use squashfs/tmpfs and aufs [13:41] like 50M or so [13:41] butu indeed that eats your ram [13:41] *but [13:41] And isn't persistent. [13:41] And isn't optimised for NAND. [13:41] you could have two partitions :) [13:42] * persia stops adding to the list [13:42] one ubi one squash [13:42] rw goes to ubi [13:42] Why not just stick a squash image in the ubifs if you want to play that way: easier. Still, no benefits from double-compression. [13:43] oh, indeed [13:43] i didnt think about stacking [14:03] someone remember why 'make' is important? [14:04] pfft, no need for make in shellscripts :P [14:05] and dash is so tiny and fast [14:05] but we got make by default [14:05] So we can build drivers if neceesary. [14:05] Some folk can't access the network until they do some DKMS magic. [14:06] Oh, and some stuff is implemented in make (it's a fine scripting language) [14:07] ogra: I'm not in charge of virtualization as a whole, no [14:07] to be safely removed: aptitude with deps - 12MB [14:08] lool, wrt arm indeed [14:08] hrw, aptitude should die die die [14:08] * lool uses aptitude every day [14:08] shudder [14:09] I would also kill console-setup/xkb-data/kbd - ~7MB [14:11] console-terminus can be sacrified if space is needed - 800KB [14:12] gnupg support another few MB [14:13] only if you use serial only [14:14] kernel has usable fonts so terminus can be dropped [14:14] tasksel is next 3.5MB [14:16] plymouth needs fixed deps to not depend on libdrm-{intel/radeon/noveuau} on ~x86(-64) [14:18] ogra: So, without aptitude, how do you do `aptitude why`? [14:49] persia, why would i do aptitude why ? [14:50] Maybe you wouldn't. I use why and why not frequenty to sort out issues each cycle. === jmcgee|gone is now known as jmcgee [15:06] never used 'why' before ;D [15:07] * Mike^ have kubuntu on omap. but it's really slow :( [15:07] heh [15:08] did you expect anything different ? [15:08] turn off all that blending :-) [15:08] ogra: I have not patience to wait untill settings will open [15:08] use netbook then :) [15:09] ogra: I used kubuntu-netbook [15:09] ubuntu-netbook indeed :) [15:10] ubuntu-netbook is much faster [15:10] hmm.. to get to <100MB would be hard [15:10] no busybox in usable package [15:11] ogra: so, the only thing left to make me completely happy is to include cm-t35 into official ubunu omap kernel ;-) [15:11] Mike^, well, if lrg gives amitk a fix for the audio issue that can happen in 10.10 [15:13] ogra, amitk said it had somthing to do with CONFIG_OMAP_MUX rather than audio... [15:13] muxing... [15:13] and, anyway, I was cheating, I had to apply two patches for headed configuration. [15:13] * XorA hides under the table away from muxing [15:14] oh, right the audio issue was AM3517-EVM [15:15] muxing is a general issue for all omap devices, but I think beagleboard should be fixed these days to work [15:15] I thought beagle has all the muxing in the bootloaders [15:16] Mike^: its also in the kernel, but for a long time was wrong [15:16] XoRA: that's why beagleboard doesn't boot with muxing enabled [15:16] and it should be fixed in kernel on all omap3/4 devices as there is no guarantee there is a bootloader [15:21] XorA: I know the MUX work, but do you remember if it went in in 2.6.34-rc? [15:22] amitk: no, I lost track [15:23] amitk: koen is normally a good person to ask about that kind of thing [15:23] ok [15:35] amitk: do you remember by chance what was the problem with beagleboard and MUX? [15:37] someone remember which package contains mk-sbuild? [15:38] ubuntu-dev-tools iirc [15:40] Yes [15:40] hrw: FYI, packages.ubuntu.com let you query this [15:40] right [15:57] * ogra fixed the screenshot on the beagle page === bjf is now known as elBoto === elBoto is now known as bjf === XorA is now known as XorA|gone [17:02] has anyone tried installing Lucid on a BeagleBoard? [17:04] inhand1, https://wiki.ubuntu.com/ARM/Beagle [17:07] orga, That is what I am working on. Have you done it? [17:07] yes [17:08] I cannot seem to get any USB devices working during the install. Therefore, no keyboard, mouse, or network. [17:09] which of the three images do you use ? [17:09] and what beagle revision [17:09] I tried all three [17:09] the images are all tested on a C4 beagle [17:09] beagle = C4 [17:09] ok [17:09] powered hub ? [17:10] yes, on either OTG or Host [17:10] (it wont work without) [17:10] OTG isnt supported yet with that kernel [17:10] ok, that helps. [17:11] but with powered hub attached to host port you should definately get kbd and mouse [17:11] can I get debug on the serial port? [17:11] for the network it depends on your NIC, apparently GrueMaster just found that moschip NICs dont work [17:12] you can, you need to edit boot.scr [17:12] i have asix and smsc LAN9514 [17:12] there is a section on the wikipage above [17:12] asix works fine for me [17:12] i just finished a netinstall test [17:12] ogra: That's only on the netboot kernel. The live image and installed image works fine. [17:12] right [17:13] the live image has all firmware available [17:13] i think the netbook image only includes a subset [17:13] i need to talk to amitk about that [17:13] *netboot [17:14] do all three use the same boot kernel? [17:15]