**** BEGIN LOGGING AT Thu Jan 21 02:59:56 2010 Jan 21 03:05:08 eggonlea: yeah, ogra might be online this afternoon, heh Jan 21 03:12:20 cooloney: thanks Jan 21 03:12:33 orga: rootstock in LP seems fine! Jan 21 03:20:45 eggonlea: excellent Jan 21 08:47:44 orga: minimal rootfs could be rootstock'ed. But met segmentation fault error with ubuntu-desktop seed. Jan 21 08:49:43 orga: log pasted at http://paste.ubuntu.com/359976/ Jan 21 08:58:04 eggonlea, hmm, that looks bad, could be an issue with the versatile kernel we use (its quite old) i plan to work on rootstock from tomorrow on and will switch it to the archive kernel then Jan 21 08:59:02 orga: got it. do you have a workable kernel at hand? Jan 21 08:59:03 did you try if it happens too if you ise a -minimal image and run apt-get install ubuntu-desktop on it later (either in VM or real HW) Jan 21 08:59:16 i dont know if the one in the archive works Jan 21 09:00:06 same Jan 21 09:00:08 I tried that. Jan 21 09:00:19 http://ports.ubuntu.com/pool/main/l/linux/linux-image-2.6.32-11-versatile_2.6.32-11.15_armel.deb Jan 21 09:01:27 let me try this. Jan 21 09:01:45 you tried what ? :) Jan 21 09:01:53 on real HW or in VM ? Jan 21 09:02:17 if it happens in VM too then its a pointer to an issue with the old versatile kernel Jan 21 09:02:54 I tried 1) rootstock -s ubuntu-desktop in VM (fail); and 2) rootstock -s minimal in VM (ok) and then apt-get install ubuntu-desktop on HW (fail). Jan 21 09:03:29 hmm, on HW points to a bigger issue thats rather not rootstock related Jan 21 09:04:22 I will try 1) rootstock -s ubuntu-desktop with your new versatile .32 kernel; and 2) apt-get install ubuntu-desktop in VM if 1) fails. Jan 21 09:05:02 but I don't think there is any actual difference between 1 and 2, right? Jan 21 09:07:51 shouldnt be Jan 21 09:08:09 but if it fails on real HW thats very likely not related to rootstock Jan 21 09:08:30 but either apt or an issue with the HW Jan 21 09:08:39 (or kernel you use on the HW) Jan 21 10:46:52 asac, any idea what we do with/without plymouth ? doesnt look like we can build it on arm in a functional way but having any kind of splash would be nice Jan 21 10:55:01 ogra: the splash solution from karmic doesnt work anymore? Jan 21 10:55:05 isnt that xsplash? Jan 21 10:55:10 usplash Jan 21 10:55:21 xsplash is for the time when X is up Jan 21 10:55:25 right Jan 21 10:55:26 ok Jan 21 10:55:34 plymouth and usplash are for the parts before that Jan 21 10:55:35 did we use usplash at all? Jan 21 10:55:39 yes Jan 21 10:55:49 in karmic (it was broken on arm in jaunty) Jan 21 10:55:50 what speaks against keeping that? Jan 21 10:55:57 nobody maintains it Jan 21 10:56:15 and i guess the team will demote it if we dont want to keep it Jan 21 10:56:46 but i see no solition for plymouth atm so we'd stay without any early splash Jan 21 10:57:55 though usplash is said to slow down the new bootprocess (we'd need to talk with Keybuk here i think) Jan 21 10:58:15 who is leading the plymouth effort? Jan 21 10:58:18 is that foundations too? Jan 21 10:58:24 Keybuk ? not sure Jan 21 10:58:32 i think mvo worked on it as well Jan 21 10:58:40 likely foundations, yes Jan 21 10:58:44 ok so its not desktop. Jan 21 11:02:50 asac, ogra, I got a patch for likewise-open, just waiting on test compilation, but hopefully I'll need a sponsor soonish :-) Jan 21 11:03:13 NCommander, good work ! Jan 21 11:03:50 ogra, it tries to determine how many processors there are from grepping /proc/cpuinfo. I just hardcoded it to 1 Jan 21 11:03:55 should make it build on all architectures Jan 21 11:04:31 asac, Bug #510585 ... not sure what to set for status etc ... i guess its rather wishlist Jan 21 11:04:32 Launchpad bug 510585 in uboot-imx "uboot-imx should have hush shell support enabled by default" [Undecided,New] https://launchpad.net/bugs/510585 Jan 21 11:05:31 NCommander, well, might make sense on x86 to use more than one CPU so a patch that only hardcodes it on !x86 would probably be better Jan 21 11:06:14 ogra, that means making the logic a bit uglier since then I need to grep uname but I'll look at it Jan 21 11:06:33 cant you use DEB_HOST_ARCH or something ? Jan 21 11:13:34 NCommander: for parallel building? Jan 21 11:13:58 asac, ogra, its not in the debian/rules file; its from the actual packages build system Jan 21 11:15:00 well, then uname might probably be better Jan 21 11:16:16 NCommander: whats the purpose? Jan 21 11:17:10 asac, ogra, it tries to determine how many processors there are from grepping /proc/cpuinfo. I just hardcoded it to 1 Jan 21 11:17:26 * NCommander kicks jocote Jan 21 11:17:52 NCommander: that doesnt explain the purpose for me ;) Jan 21 11:17:54 anyway Jan 21 11:18:00 give me patch Jan 21 11:18:11 asac, speedier builds on SMP machines :) Jan 21 11:18:13 * NCommander doesn't get why the version number is likewise-open_5.4.0.39949-3 if there's no Debian version :-/ Jan 21 11:18:29 ask the server team Jan 21 11:18:32 asac, I need to upload to a PPA. jocote has an unclear chroot which I think triggered an FTBFS Jan 21 12:04:19 Hallo Leute :-) ich suche ein Ubuntu arm Toolchain-Howto für 2.4.18-rmk3 armv4l chip arm720t (Owasys-Gerät). Kann mir da jemand weiterhelfen? Ich muss dropbear und appweb portieren und habe arm-linux-gcc als tar bekommen. Einfachen c-code kann ich compilieren. Zur Zeit versuche ich mich an appweb. Würde mich über Tips freuen. Jan 21 12:12:29 Noisi, hey, this is an english speaking channel :) Jan 21 12:15:52 ok sorry :-) Jan 21 13:09:54 grmpf ... adding the script support to the uboot defaults is harder than i though Jan 21 13:09:58 t Jan 21 13:14:33 hmm, a "break" command would be so handy Jan 21 13:14:47 hush shell is so lame Jan 21 14:15:22 GRRR Jan 21 14:20:40 * NCommander kicks likewise Jan 21 14:20:43 Freaking configure checks Jan 21 15:16:21 dmart: plars: lucid chromoium with full thumb and neon is here: https://edge.launchpad.net/~ubuntu-mozilla-security/+archive/nss3.12.3 Jan 21 15:16:26 chromium-browser - 4.0.303.0~svn20100120r36627-0ubuntu1~ucd1+asac2 Jan 21 15:17:04 uploading firefox 3.6 all-static now (had to get quota bumped first) Jan 21 15:17:11 asac: cool :) Jan 21 15:18:18 let me know if it starts on imx51 Jan 21 15:18:24 thumb2 actually Jan 21 15:20:22 i sawved the lucid builds without NEON/thumb2 here: http://people.canonical.com/~asac/tmp/chromium-armel-lucid/ Jan 21 15:20:53 asac, is that supposed to work on non neon HW ? Jan 21 15:21:02 ogra: no. Jan 21 15:21:04 i.e. b2.5 Jan 21 15:21:08 ogra: thats a benchmark thing Jan 21 15:21:13 ah, k Jan 21 15:21:17 ogra: b2.5 most likely works Jan 21 15:21:35 well, i'll try if i ever get my board booted again :P Jan 21 15:21:38 dmart did some tests and the CPU hangs are not really happen frequently Jan 21 15:22:13 i just wanted to use the chance that we have a NEON kernel to check if there is a difference Jan 21 15:22:14 * ogra only saw the uboot prompt since getting up today ... Jan 21 15:22:21 hehe Jan 21 15:22:24 i know that feeling Jan 21 15:22:27 endless hangs ;) Jan 21 15:22:31 but i think i got it now Jan 21 15:22:35 are the best :-P Jan 21 15:22:36 no, no hangs Jan 21 15:22:43 yes, thats good ;) Jan 21 15:22:53 just figuring out a logic that doesnt make hush hsell laugh at me Jan 21 15:23:00 hehe Jan 21 15:23:05 but i think i got it now ... Jan 21 15:23:44 if mmc is detected it will try to load boot.scr ... if boot.scr isnt there it will try to load reamdisk ... if ramdisk isnt there it will try to load kernel ... Jan 21 15:23:48 can we move debian-cd to uboot wihtout hushshell as a first step? Jan 21 15:23:49 then it will boot Jan 21 15:24:06 if mmc isnt there it will automatically exec the netboot commands Jan 21 15:24:29 just thinking it would be a good thing for a peer to get started on that spec by writing good hushshell script Jan 21 15:24:34 i'll also try to fall back on netboot if kernel isnt found Jan 21 15:24:39 and you could work on better stuff ;) Jan 21 15:24:52 well, i'm done ... just need to test it Jan 21 15:24:55 ok Jan 21 15:24:59 * ogra waits for a build to finish Jan 21 15:25:03 when can this go into debian-cd? Jan 21 15:25:38 well, debian-cd will mainly get the uboot-image-script ported Jan 21 15:25:51 i just need to add boot.scr generation once uboot works Jan 21 15:26:30 but thats mainly four to five lines ... Jan 21 15:27:36 ogra: how about producing the universal boot.scr we aim at in the uboot-imx package? Jan 21 15:27:38 the hard part was to patch uboot to actuaqlly do something with boot.scr in an intelligent way Jan 21 15:27:50 not for now, but in long run i dont see why we need to do that in debian-cd Jan 21 15:27:51 that wont work Jan 21 15:27:54 why? Jan 21 15:28:02 boot.scr will have the cmdline Jan 21 15:28:07 ok Jan 21 15:28:17 for live images thats hardcoded, for installs it needs the UUID Jan 21 15:28:39 we need to generate/change boot.scr from flash-kernel or d-i Jan 21 15:29:33 but that code is already there thanks to NCommander and lool, thats something we only need to copy Jan 21 15:30:13 ogra, you can probably just use the same extact code in flash-kernel(-installer), and minor modifications to the code in d-cd Jan 21 15:30:37 NCommander, yes for flash-kernel, no for d-cd Jan 21 15:31:03 our uboot is way cleverer now than the dove one ... we need a lot less code in d-cd Jan 21 15:32:15 i added the iteration over filesystems and partitions directly into uboot, no need for complex boot.scr stuff anymore Jan 21 15:32:38 ogra, we should port that code to dove Jan 21 15:32:50 its just changes to the default env Jan 21 15:33:08 on dove it would likely be more complex though, since you have ide and usb too Jan 21 15:33:09 ogra, thats probably the sanest way to do so, then you can check the variables and iterate as necessary Jan 21 15:33:26 babbage only has mmc Jan 21 15:33:41 but thats only one more loop Jan 21 15:33:45 ogra, right Jan 21 15:33:48 and some inits Jan 21 15:37:54 perfect ! Jan 21 15:38:24 no boot.scr gets me to busybox (because we dont have a default rootdevice set atm) Jan 21 15:38:40 * ogra deletes uinitrd Jan 21 15:39:47 and no initrd just boots the kernel :) Jan 21 15:41:51 so users only have to set cmdline at the prompt and run "boot" if they want to tinker manually :) Jan 21 15:41:59 the rest happens automatically Jan 21 15:42:08 now lets try a boot.scr Jan 21 15:44:06 GOSH ! Jan 21 15:44:09 it WORKS ! Jan 21 15:44:14 i'm so awesome !!! Jan 21 15:44:23 * ogra dances Jan 21 15:49:42 ogra: you are awesome indeed Jan 21 15:49:51 now get this plumbered into our images ;) Jan 21 15:49:52 hehe Jan 21 15:49:55 ok ttyl Jan 21 15:50:00 have to commute to hamburg Jan 21 15:50:05 bbl Jan 21 15:55:25 * ogra will upload in a minute Jan 21 15:58:48 ogra, very nice. I think I need to merge that back into Dove :-) Jan 21 15:59:25 NCommander, its not much different from what you do in your boot.scr Jan 21 15:59:51 i just use it in the defaults here Jan 21 16:01:25 NCommander, http://paste.ubuntu.com/360126/ Jan 21 16:02:24 i'll modify it a little bit to actually have bootargs_mmc Jan 21 16:03:06 (pointing to /dev/mmcblk0p2 by default) Jan 21 16:05:12 ogra, I thought you got the partition numbers set in the environment via the mmcinit command (and assiocated friends) Jan 21 16:05:39 from where should i get them ? Jan 21 16:05:59 i dont actually know on which partition the kernel or boot.scr lives Jan 21 16:06:48 the prob on imx is that we cant save anything Jan 21 16:06:59 so the defaults need to be as dynamically as possible Jan 21 16:07:10 saved settings need to live in boot.scr Jan 21 17:25:52 asac: chromium from that archive running fine for me on babbage3 Jan 21 17:25:56 asac: about to try ff3.6 Jan 21 17:26:19 fine and fast or just fine ? Jan 21 17:26:20 :) Jan 21 17:27:04 ogra: I don't have a good way of benchmarking it, but seem decently quick for the platform (expected somewhat due to lighter weight browser though) Jan 21 17:27:32 ogra: he seemed interested in whether it comes up at all though Jan 21 17:27:46 well, i think even the non neon/thumb2 version feels a lot faster than karmics FF Jan 21 17:32:08 asac, ping Jan 21 17:32:58 dmart, he is on a train atm Jan 21 17:33:13 Thought he might have gone... Jan 21 17:33:25 Do you know whether the static build of ff3.6 was uploaded yet? Jan 21 17:33:43 no, sadly not Jan 21 17:33:51 (knowing) Jan 21 17:34:20 I have no babbage3 to test the chromium rebuild yet, unfortunately. Jan 21 17:35:29 yeah, mine hasnt grown up either yet ... it was still a 2.5 last i checked :) Jan 21 17:37:09 I will try the NEON-ised version later anyway, but it might not work... Jan 21 17:37:19 he said it might Jan 21 17:37:23 https://edge.launchpad.net/~ubuntu-mozilla-security/+archive/nss3.12.3 has a FF3.6 build that he made available on the 12th, I think he's planning on putting another one there Jan 21 17:37:41 asac, is that supposed to work on non neon HW ? Jan 21 17:37:41 ogra: no. Jan 21 17:37:41 i.e. b2.5 Jan 21 17:37:43 ogra: b2.5 most likely works Jan 21 17:39:06 b2.5 may not work (but I don't have one to test on) Jan 21 17:40:45 yes, thats what i thought too Jan 21 17:40:48 ogra: just rough guessing.. launching both the 3.6 that's there and the chromium build and pointing at www.ubuntu.com (after a few cache warm-ups of course) the chromium seems to get there in about 6 seconds, FF takes about 10 Jan 21 17:41:02 yeah Jan 21 17:48:14 That's the same kind of result I was seeing Jan 21 18:26:52 dmart: the static build was uploaded right before i left Jan 21 18:26:58 had quota issues before Jan 21 18:27:02 so it got rejected yesterday Jan 21 18:27:03 let me check Jan 21 18:27:53 dmart: does the NEON/thumb2 chromium build work well? Jan 21 18:28:02 dmart: https://edge.launchpad.net/~ubuntu-mozilla-security/+archive/nss3.12.3/+build/1459774 ... ffox still spinning Jan 21 18:28:08 i think will take another hour or so Jan 21 18:28:37 I was having board problems and wasn't able to test the new Chromium build yet... also, I have no babbage 3 so it might not work, but I will try. Jan 21 18:28:48 This will probably be tomorrow now... Jan 21 18:31:00 Got to disappear now, sorry Jan 21 18:36:21 asac: chromium build is working pretty well for me so far Jan 21 18:45:43 nice Jan 21 18:45:51 plars: is that on TO3 Jan 21 18:45:52 ? Jan 21 18:46:23 asac: yes Jan 21 18:46:52 have you checked if it hangs up a TO2 or 2.5? Jan 21 18:50:37 asac: no, but I can pull out the 2.5 and try it if you'd like Jan 21 18:52:34 dmart: still there? Jan 21 18:52:50 dmart: one thing mike mentioned up front is that we should disable disk_cache in ffox 3.6 for benching Jan 21 18:53:14 dmart: so in about:config set browser.cache.disk.enable to false Jan 21 18:53:23 we probably should have that by default for arm Jan 21 18:54:04 i will shoot you a mail once ffox 3.6 has finished, so you can check that Jan 21 18:54:46 plars: would be interesting to see if chromium triggers the problematic NEON stuff Jan 21 18:58:47 asac: will try it in a bit Jan 21 19:09:55 asac: nothing extensive tried, but just opening, visiting a few sites, that chromium build is also stable for me with TO3 Jan 21 19:10:03 asac: err. 2.5 too :) Jan 21 20:07:40 plars: ok, maybe use it a bit longer. good news. Jan 21 20:07:41 thanks Jan 21 20:07:56 * asac hopes that upstream didnt decouple neon from armv7=1 **** ENDING LOGGING AT Fri Jan 22 02:59:57 2010