**** BEGIN LOGGING AT Wed Oct 03 02:59:59 2012 Oct 03 05:44:12 so what was the result of the whole discussion with srwarren? are the r16 libs getting fixed? Oct 03 08:27:22 lilstevie: it will be fixed, but maybe not in time Oct 03 08:33:41 just a question about makefile : I export a CFLAGS variable for a target that configure gstreamer. The configure step is passes but when compiling I get a bunch of errors. What I found is that at the compile step the CFLAGS have been passed with the "..." so that compiler (libtool) that this as a file. When I call the configure the same way in the shell everything is ok. So how do we avoid such export error in makefile ? Oct 03 08:36:55 infinity: but kernel installation are not overriden, right ? So after updating the new kernel will be selected. If I want to fallback I'll just have to update the symlink to uImage (BB xM) Oct 03 08:38:22 bizulk: for your makefile question, do you have a log you can share? Oct 03 08:38:44 dmart: Yes give 2 minutes to make a pastbin Oct 03 08:47:31 marvin24, hm ok Oct 03 08:47:41 marvin24, is there a package that works with precise though? Oct 03 08:54:53 dmart: http://pastebin.com/GS5W1sCZ Oct 03 08:55:42 dmart: I tried $() ${} , always getting error at compile time Oct 03 09:00:45 bizulk: Hmmm, I'm not sure exactly why that ends up quoted Oct 03 09:01:06 bizulk: What happens if you put the flags in the environment instead of on the configure command line Oct 03 09:01:09 bizulk: i.e. Oct 03 09:01:41 well I'll try it now Oct 03 09:02:48 bizulk: hang on, forget that for a moment -- do you have the Makefile? I'm wondering whether it adds inappropriate quoting to the make variables corresponding to GSTREAMER_CFLAGS and GSTREAMER_LDFLAGS Oct 03 09:03:38 well it does not work Oct 03 09:04:07 You mean the gstreamer Makefile ? Oct 03 09:05:20 bizulk: yes... or were the lines at the top of your log part of the Makefile? I had assumed they were shell commands issues before typing make Oct 03 09:05:33 I'll post the whole makefile Oct 03 09:07:02 http://pastebin.com/KaKeasVU Oct 03 09:07:43 maybe I am messing the configure step with my GLIB custom makefile. I trying to make a diff Oct 03 09:10:43 bizulk: Normally people do not put quotes directly in Makefile variable assignments, because Make doesn't need them and they become part of the variable Oct 03 09:11:37 dmart: you mean when I export ? I thought I needed them because of space Oct 03 09:15:16 lilstevie, i think i'll upload a package with the broken libs, the direver works fine without them and we can then SRU the fixed libs Oct 03 09:17:01 bizulk: Make generally doesn't use quotes at all Oct 03 09:17:28 dmart: you're right I've just been inspecting the gstreamer generated Makefile Oct 03 09:17:30 bizulk: but rule commands may need them because those are executed by the shell Oct 03 09:17:38 Maybe something like this: Oct 03 09:17:39 http://pastebin.ubuntu.com/1257570/ Oct 03 09:17:40 ogra_, ok cool, I do find I care more about the driver than the libs anyway Oct 03 09:18:17 yeah, not many gles apps around yet and unity is broken with the nvidia driver anyway Oct 03 09:18:24 dmart: right. I got it working now but I don't know why. So I'll use your models Oct 03 09:19:06 ogra_, even with r16? Oct 03 09:19:29 infinity, would you let the tegar package in even with broken libs (to make SRUing the fix easier) ? Oct 03 09:20:01 lilstevie, apparently, though the reports i had were from precise ... with luck quantal unity works better (though i doubt it) Oct 03 09:20:49 shame Oct 03 09:21:06 dmart: ok it seems to be compiling. Variable assignement rule is using the whole line after ' = ' and not quote then. Oct 03 09:21:20 the patches should be largely the same in both, just that they are upstream in quantal Oct 03 09:21:20 yes Oct 03 09:21:32 you would have thought they would have fixed it though, seeing as their test env is now a 12.04 armhf image Oct 03 09:21:49 which still falls back to unity2d Oct 03 09:22:13 when i asked them last they were convinced unity works ... turned out in the end that they looked at 2d :) Oct 03 09:22:13 yeah, but it doesn't fall back with the tegra driver (although it might with broken libs) Oct 03 09:22:41 yeah. not anymore Oct 03 09:24:32 hm Oct 03 09:24:40 dmart: Thanks for help Oct 03 09:25:07 I'm not quite satisfied with our r16 kernel anyway Oct 03 09:25:18 bizulk: no problem Oct 03 11:21:20 janimo, ! Oct 03 11:21:40 janimo, could it be that you changed the config wrt usb-storage in the last kernel ? Oct 03 11:21:58 people arent finding the USB key during install anymore Oct 03 11:23:25 ogra_, you mean the very last kernel which was supposed to be the NVAVP change? Oct 03 11:24:05 janimo, dunno, is that in already (meta ?) whartever is on todays image doesnt find a /dev/sda1 anymore in the installer initrd Oct 03 11:24:19 USB plug events are seen Oct 03 11:24:37 that smells like usb-storage moved from =y to =m Oct 03 11:25:41 (the initrd is built without any modules due to size issues if you remember) Oct 03 11:26:01 sure. I do not remember touching usb-storage though Oct 03 11:26:03 I'll check Oct 03 11:26:21 yeah, might be marvin24 changed it Oct 03 11:26:35 that would break everyone who runs rootfs on USB though Oct 03 11:26:53 I do not use defconfig so not sure how marvi could have affected this though Oct 03 11:27:00 ah, k Oct 03 11:27:27 but yes, thatwas it. storage is modular :( Oct 03 11:27:38 heh Oct 03 11:27:43 part of the first r16 kernel upload Oct 03 11:27:50 where I actually did merge with defconfig Oct 03 11:27:59 I hate this crap (kernel packaging) Oct 03 11:28:01 funny, i thought that made it onto beta2 Oct 03 11:28:09 you wanted it :) Oct 03 11:28:25 I was young and stupid Oct 03 11:28:42 i wuld have kept it (and maintained it in tarballs) ... now its in git .... else i'd take it back :P Oct 03 11:29:11 I think it is better for it to be in git, as it is - in theory at least - easier to team-maintain Oct 03 11:29:20 right now the lack of interest not of git is the blocker Oct 03 11:29:51 ogra_, but really git is winning , sooner or later you will have to use it Oct 03 11:29:56 so why not now ;) Oct 03 11:30:08 enough other stuff on my plate :) Oct 03 11:30:28 i'm happy if i can git pull, the rest i stay away from Oct 03 11:32:01 janimo, ogra_, sorry folks that I broke it Oct 03 11:32:04 well it has an annoying hurdle at the beginning of the learning curve, but it pays off later Oct 03 11:32:36 marvin24, np, this is error-prone and boring stuff Oct 03 11:32:46 pc distros prefer all modules Oct 03 11:32:54 and load them in the initrd stage Oct 03 11:33:08 I wonder why this doesn't work on ARM Oct 03 11:33:31 it does Oct 03 11:33:33 just fine Oct 03 11:33:36 it's not ARM but our ac100 initrd Oct 03 11:33:47 its just that there are no modules to load Oct 03 11:33:58 since we dont include any in the initrd on ac100 Oct 03 11:34:07 else it would get to big Oct 03 11:34:16 people are just too lazy in ARM world to do things in the generic way and just take vendor trees and build everything in the kernel Oct 03 11:34:35 suihkulokki, its not that :) Oct 03 11:34:43 it is :P Oct 03 11:34:49 its a constraing of the dveices bootloader we have to obey to Oct 03 11:35:00 but it is changing with the unified kernels and device trees etc Oct 03 11:35:03 your initrd cant be bigger than 2.5M or so Oct 03 11:35:26 or even 2M ... i dont remember the exact value Oct 03 11:36:12 if i want to use a standard ubuntu initrd i have to cut down size massively due to setting other defaults for initramfs-tools Oct 03 11:36:46 ogra@panda:~$ ls -lh /boot/initrd.img-* Oct 03 11:36:46 -rw-r--r-- 1 root root 4.6M Sep 26 17:37 /boot/initrd.img-3.5.0-211-omap4 Oct 03 11:36:46 ogra@panda:~$ Oct 03 11:36:53 that would be the normal size Oct 03 11:37:09 no way to fit that into the ac100 Oct 03 11:39:28 s7due to/through/ Oct 03 11:41:43 janimo, oh, you are not in #arm anymore on the other server :( Oct 03 11:42:36 ogra_, oh that's because that channel used to say 'closing. move along' or similar half a year ago Oct 03 11:42:39 and it was inactive Oct 03 11:42:52 did I miss something lately? Oct 03 11:43:13 well, the channel has still 35 residents :) Oct 03 11:43:24 ogra_, but are they alive? Oct 03 11:43:25 and we occasionally talk there Oct 03 11:43:32 rarely :) Oct 03 11:43:57 i just didnt want to talk in #canonical ... and you arent in#distro either Oct 03 11:44:17 ogra_, well, I thought those too were kind of quiet Oct 03 12:15:15 janimo: it seems you were right a few weeks ago ... Oct 03 12:15:35 there is significant work towards multi arch kernel on arm Oct 03 12:15:48 or better multi cpu Oct 03 12:16:00 I wonder how they will solve the errata problem Oct 03 12:16:49 devicetree needs to learn about them Oct 03 12:16:57 marvin24, tegra in 3.7 seems quite nice (from a codebase point of view, not sure how it actually works) Oct 03 12:17:29 ogra_, but still checking for them incurs runtime overhead even for unaffected hw Oct 03 12:17:42 yes, seems tegra is the first platform to kill board files Oct 03 12:17:55 unless some live patching is done at bootup, which may not be that hard on ARM as it is on x86. No idea Oct 03 12:18:50 janimo, you dont check for anything thats not in your DT database for the HW you run on Oct 03 12:18:56 thats the purpose of DT Oct 03 12:19:13 ogra_, I mean errata code is inside #ifdef Oct 03 12:19:23 and not built unless selected for the particular target in config Oct 03 12:19:28 that needs changing indeed Oct 03 12:19:40 so if you want to support all boards you make the checks at runtime Oct 03 12:19:44 so some overhwead Oct 03 12:19:54 which would be a part of "devicetree needs to learn about them" Oct 03 12:20:24 or decree you go one stepo further and ARM kernels will not only not accept machines that are not DTbaserd, but not accept machines that have hardware bugs Oct 03 12:20:31 would make arm kernel maintenance so much easier Oct 03 13:40:11 I am compiling glibc against arm2009q1 (code sourcery), and gstreamer, for BB xM. But when calling for gst-inspect I have this error : "/lib/libglib-2.0.so.0: /lib/libc.so.6: version `GLIBC_2.9' not found (required by /lib/libglib-2.0.so.0)". I can't understand as I did install the libc from the toolchain dir. Oct 03 13:41:10 How can glib ask for a glibc version later than the one it has been compiled with (what a headache !) Oct 03 13:41:44 ugh, why do you use such an outdated toolchain Oct 03 13:45:17 I was using the last one but because I could have tidspbridge support (still not sure about the reason) I fallback to the one my colleagues are actually using Oct 03 13:45:48 Actually the glib version is "glib-2.24.1" Oct 03 13:46:36 why did I choose this one ? Maybe because I wanted to use same gstreamer version as the gst-dsp project and that it is compiled with this version og glib Oct 03 13:49:49 I'll cry if I have to recompile the whole stuff (BSP and userspace) with another toolchain Oct 03 14:06:31 bizulk, you recompiled ubuntu ? Oct 03 14:07:05 note that the whole ubuntu userspace is compiled with gcc 4.7 Oct 03 14:07:28 i_m not even sure that old toolchain you use has proper armhf support Oct 03 14:07:50 no that standalone stuff (I have to) , I am using ubuntu as "witness test". I'm wondering If I can generate minimal FS with ubuntu Oct 03 14:08:13 use debootstrap Oct 03 14:08:28 or qemu-debootstrap for cross building a chroot/rootfs Oct 03 14:08:58 (note that this will be compelertly unconfigured though) Oct 03 14:17:24 can anyone help me with Qt's audio classes under ubuntu for pandaboard Oct 03 14:17:44 ogra_: I'll take an eye one it. infortunately I have to be base on the one I receive and populate it with the builds I make Oct 03 14:44:58 ok I recompile with a more recent kernel arm-2012.03-57-lite Oct 03 14:45:22 s/kernel/compiler/ ? Oct 03 14:45:23 I hope one day we'll adopt OE/Ubuntu u Oct 03 14:45:28 compiler Oct 03 14:45:35 urgs Oct 03 14:45:48 thats like saying for the desktop you want a merge of gentoo and ubuntu Oct 03 14:46:15 ubuntu isnt an embedded OS ... dont use it as one Oct 03 14:46:35 so why port on a omap ? Oct 03 14:46:53 we needed to start somewhere Oct 03 14:47:21 ompa support is a legacy we only carry on because we get it for free from te mainline kernel Oct 03 14:47:25 *omap Oct 03 14:47:41 focus in on cortex-a9 and beyond Oct 03 14:48:06 which are systems that are capable of running a general purpose desktop distro Oct 03 14:51:33 bug 1028905 Oct 03 14:51:34 Launchpad bug 1028905 in debian-installer-utils "cdrom-detect in quantal omap4 hangs trying to look for install media on an extended partition" [High,Fix released] https://launchpad.net/bugs/1028905 Oct 03 16:51:50 janimo, did you include the fuse fix as well in your ac100 upload (changelog doesnt talk about it ) Oct 03 16:52:23 ogra_, I rebased on marvin's branch so it should be there Oct 03 16:52:30 good Oct 03 16:52:32 I have a LP buglink which should point to that Oct 03 16:52:37 just wanted to make sure Oct 03 16:52:39 but indeed, forgot to mention it otherwise Oct 03 16:52:55 I am not sure of anything now, so let's see who yells when it builds :) Oct 03 16:52:56 hopefully thats the last upload :) Oct 03 16:53:03 yeah Oct 03 16:53:12 theoretically it should be good now Oct 03 16:54:14 Can I just yell for the sake of yelling? Oct 03 16:54:43 * ogra_ plugs his ears Oct 03 16:54:44 infinity, always Oct 03 16:54:46 GO ON ! Oct 03 16:54:59 :) Oct 03 16:55:04 Well, that's no fun. Oct 03 16:57:31 dannf: Around? Oct 03 16:57:40 infinity: yeah Oct 03 16:57:59 dannf: Can you re-do your qemu-linaro SRU with the patch headers actually being sane? :P Oct 03 16:58:19 dannf: They include ufilled boilerplate, a reference to an unrelated bug, etc. Oct 03 16:58:38 * dannf looks Oct 03 16:59:02 dannf: Err, oh, I see the problem. There's a debian-changes patch in there. Oct 03 16:59:21 dannf: So, yeah. Fix that, please. :P Oct 03 16:59:59 dannf: That debian-changes patch looks like maybe a mistake while fiddling? Can't see why you'd want to remove the stdint.h include. Oct 03 17:00:29 dannf: http://launchpadlibrarian.net/117670218/qemu-linaro_1.0.50-2012.03-0ubuntu2_1.0.50-2012.03-0ubuntu2.1.diff.gz Oct 03 17:00:39 dannf: (Always pays to debdiff and cruft-check before upload) Oct 03 17:00:54 debdiff ++ Oct 03 17:00:56 dannf: Rejecting that one, give me something gooder. Oct 03 17:02:09 infinity: ok - i don't see those changes in my branch.. did i post this debdiff? Oct 03 17:02:27 dannf: That's the diff generated in the queue from the package you uploaded... Oct 03 17:02:36 i didn't upload a package Oct 03 17:02:44 Oh. Did someone else sponsor that? Oct 03 17:02:49 except to my ppa (i don't have ubuntu upload rights) Oct 03 17:03:02 here's my changes: https://code.launchpad.net/~dannf/qemu-linaro/highbank Oct 03 17:03:44 * dannf remembers hallyn poking me last week while i was travelling about qemu-linaro - i'd forgotten about that & never got back to him Oct 03 17:04:39 dannf: Check, I'm chastising the wrong person. I didn't check the signature to see who sponsored it. :P Oct 03 17:04:57 infinity: np - thanks for working on this Oct 03 17:05:36 Looks like it was mvo who broke it. That's surprising. Oct 03 17:05:41 I so wanted it to be you. Oct 03 17:05:44 *sniff* Oct 03 17:05:56 I'll re-do this and upload. Oct 03 17:07:22 heh - sometimes i wonder if i should try and get upload rights.. then i remember the blame that comes w/ privilege :) Oct 03 17:09:32 Oh, curious, mvo's debian-changes patch unapplied Colin's FTBFS fix. Oct 03 17:09:34 Weird. Oct 03 17:09:42 Aaaanyhow. Reuploading a less broken version. Oct 03 17:26:20 dannf: Alright, fixed version uploaded, double-checked that the diffs are now clean, and accepted into -proposed. Oct 03 17:27:26 yay! Oct 03 18:38:33 * xnox is playing with ac100 android install before reinstalling it with ubuntu Oct 03 19:29:27 * xnox ac100 no tarball found hm =/ Oct 03 21:44:10 xnox: Why would you reinstall that piece of vile vomit back on the AC100? It isn't even true Android, but a hackified version (similar to the nook, only not as nice). Oct 03 21:46:49 lolwhat Oct 03 21:49:47 fly-away: The Toshiba AC100 came preloaded with a heavily modified version of Android (2.3 iirc). Very ugly and not as easy to use as true Android. Oct 03 21:50:13 The system could have been very good if it had more memory and a real OS out of the box. Oct 03 21:51:04 that preinstalled Android pretty close to original Oct 03 21:51:33 one annoying thing - android is for touchscreens now Oct 03 21:51:53 keyboard really unusable Oct 03 21:53:36 It felt a lot different than the android I worked with on other systems. At any rate, there was a lot missing from that system, both hardware and software. Oct 03 21:53:52 well that version of android is pretty old and had "hacks" for keyboard/mouse support Oct 03 21:54:44 true. Oct 03 21:56:48 in that android video plays with subtitles very smoothly Oct 03 21:56:55 if you remember) **** ENDING LOGGING AT Thu Oct 04 02:59:58 2012