**** BEGIN LOGGING AT Mon Apr 18 02:59:57 2011 Apr 18 06:34:46 ogra_: hi Apr 18 06:35:39 ogra_: have you seen the messages on pandaboard google groups? some folks report that headless image don't work with 8Gb or 16Gb, but work with 4Gb? Apr 18 09:14:02 ndec, whats the bug number? and did they attach the jasper.log ? Apr 18 09:14:54 * ogra_ hasnt heard of any errors with 8/16G cards yet Apr 18 09:25:09 qemu + u-boot = fail Apr 18 09:28:50 ppisati: Nah Apr 18 09:28:55 ppisati: Which u-boot? Apr 18 11:04:47 ogra_, janimo: shrink-wrap came up again on today's linaro call. looks like it should be disabled for the natty release, or at least for -updates Apr 18 11:05:55 shrink-wrap ? Apr 18 11:05:58 doko: michaelh submitted a branch to disable it by default Apr 18 11:06:04 doko: In gcc-linaro Apr 18 11:06:14 doko: Not sure you will have the chance to integrate this though Apr 18 11:06:31 https://code.launchpad.net/~michaelh1/gcc-linaro/disable-shrink-wrap/+merge/58055 Apr 18 11:07:56 lool: it's already in gcc-4.5 for all archs but armel Apr 18 11:09:33 doko: Okay Apr 18 11:11:00 doko, well, throw it in if you can, if you cant throw it in updates :) Apr 18 11:11:21 (as if we had a choice :P just do it :) ) Apr 18 11:21:20 doko, sounds fine Apr 18 11:21:49 doko, not sure what other packages are affected and will remain so without a rebuild Apr 18 11:24:27 janimo: right, but you can't rebuild everything :-/ Apr 18 11:25:01 sure, disabling is better then leaving the optimization on, I was just wondering aloud :) Apr 18 11:25:18 we can rebuild in -updates if needed i think Apr 18 11:25:31 in case we run into other issues Apr 18 11:27:10 sure. The more serious issues are identified as bugs already, the remaining ones are less used universe packages Apr 18 11:27:19 yeah Apr 18 11:27:28 btw, what was with that banshee bug ? Apr 18 11:27:42 tobin said we need to disable something to make it eat less cpu Apr 18 11:27:49 is anyone on top of that ? Apr 18 11:28:01 ogra_, no idea re banshee Apr 18 11:28:05 hmm Apr 18 11:28:11 * ogra_ digs his bugmail Apr 18 11:28:21 I remember rsalveti noticing much higher CPU usage than with rhythmbox and mpg123 Apr 18 11:28:28 but I have no other info Apr 18 11:28:46 20% 10% and 3% CPU IIRC Apr 18 11:28:58 yeah Apr 18 11:29:12 there was an option that you can disable but i cant find a bug Apr 18 11:29:36 I know!! Let's allow mono to take advantage of dual cores. It should make that CPU usage better :D Apr 18 11:30:02 disable what? Does it do something extra for mp3 which is optional? Apr 18 11:30:13 does it not use gstreamer? Apr 18 11:30:36 there is a lib or something you can disable that significantly drops the cpu usage Apr 18 11:30:56 lets wait for GrueMaster or rsalveti, i think they already looked into it Apr 18 11:30:59 LDFLAGS -= -lbloat Apr 18 11:31:05 lol Apr 18 11:31:20 the discussion sounded more like a runtime thing Apr 18 11:31:51 banshee has a lot of plugins linked in (lots of configure options anyway), could be one of those Apr 18 11:31:59 yep Apr 18 11:33:33 * ogra_ goes back to bang his head against pulse Apr 18 11:33:59 life was so easy when we only had esound :P Apr 18 11:53:31 esound? I had artsd Apr 18 12:12:49 * ogra_ curses Apr 18 12:13:14 so adding the one missing patch to pulse breaks everything again Apr 18 12:14:51 thats just depressing Apr 18 12:22:01 ogra_: no LP report, just discussion on google group: http://groups.google.com/group/pandaboard/browse_thread/thread/c2287d603dc325e0 Apr 18 12:26:58 ndec, well, i cant do much without logs, we didnt have any such errors when testing with various cards Apr 18 12:27:45 ogra_: ok. i just wanted to make sure you've seen this. at least 2 persons have had this problem and reported on the panda mailing list. Apr 18 12:28:02 that guy definitely trashed it by rebooting in the middle of the resize process Apr 18 12:29:44 ogra_: well he said he waited 1 hour before rebooting Apr 18 12:31:57 well, i need the jasper.log, cant say anything without Apr 18 12:32:24 it might well be that there was a regression in the dailies, he should definitely try the beta2 instead Apr 18 12:37:02 * ogra_ sighs ... Apr 18 12:37:12 really doesnt look like that sound stuff will work at all Apr 18 12:37:14 :( Apr 18 12:47:44 ndec, do you know if your sound people actually use the ubuntu kernel ? Apr 18 12:48:39 ogra_: yes. at least this is what i ask... Apr 18 12:48:50 weird Apr 18 12:49:52 * ogra_ doesnt get why it doesnt work then Apr 18 13:11:09 ndec: I took today's OMAP preinstalled image and dd-ed it to three files of sizes 2 G, 4 G and 8 G, and started qemu-linaro with -M beaglexm -m 512 -sd on all three; both the 2G and 4G instances booted up to d-i (after resizing and rebooting), I'm waiting on the 8G still (no error so far); this is not enough to conclude that there is no problem though, as physical cards don't have exact multiples of Gs, but relatively random sizes Apr 18 13:14:45 lool: i am sure the images work most of the times... my point was just to highlight to ogra_ that sometimes they don't... that might very well be 'local' issues.. Apr 18 13:15:07 ndec: it might be local, or it might be partitioning logic with specific input sizes Apr 18 13:15:13 well, it might take longer on bigger cards Apr 18 13:15:41 i must admit that i didnt test with more than 8G though my 8G card behaved fine Apr 18 13:16:02 * ogra_ hopes btrfs images in oneric will make our life easier here Apr 18 13:16:14 You decided to go with btrfs? Apr 18 13:16:21 we will investigate Apr 18 13:16:30 Did you try it out so far? Apr 18 13:16:42 afaik the big probs left wrt btrfs are bootloaders Apr 18 13:16:49 which doesnt affect us Apr 18 13:17:02 lool, not yet, but we have spec plans for it Apr 18 13:17:18 * ogra_ is to busy with release to try out new stuff atm Apr 18 13:17:26 I wouldn't have too high hopes that it will help Apr 18 13:17:42 Performance on a panda + btrfs rootfs has been terrible here, much worse than extN Apr 18 13:17:45 its like a deja vu ... last release i spent the last weeks on omap4 sound ... this release too Apr 18 13:18:02 and I had to fix some issues in natty before that would even boot without error Apr 18 13:18:08 lool, even if you aling properly and use the ssd option ? Apr 18 13:18:13 Yes Apr 18 13:18:13 *align Apr 18 13:18:16 hrm Apr 18 13:18:29 i would love nilfs2 ... but it has no resize function yet Apr 18 13:18:51 so far it was far beyond all other FSes on SD in all comparisons i have seen Apr 18 13:19:32 on the ac100 it seems to double up the access speed Apr 18 13:31:56 ndec: FTR, the 8 G image also booted up to the config screen Apr 18 13:32:42 Now it would be interesting to try with uneven sizes, but QEMU might not be ideal for this Apr 18 13:34:24 morning Apr 18 13:36:07 ogra: janimo: bug 760902 Apr 18 13:36:09 Launchpad bug 760902 in banshee "Banshee's Library Watcher should be disabled by default in Ubuntu 11.04" [Wishlist,Incomplete] https://launchpad.net/bugs/760902 Apr 18 13:36:24 kexec is really broken :( Apr 18 13:37:07 rsalveti, so it is not arm specific Apr 18 13:37:13 nops Apr 18 13:39:54 ah, i thought you only saw it on arm Apr 18 14:39:19 * ogra_ goes for his 3rd pulseaudio build today Apr 18 14:45:35 lol Apr 18 14:54:30 Morning. I see rsalveti posted the banshee bug. I actually didn't see much improvement w/o the libarary watcher, but I was only able to test on my beagleXM. Apr 18 14:55:12 GrueMaster, well, its fine, its not armel alone so i dont care much Apr 18 14:55:24 i thought it was arch specific Apr 18 15:28:05 rsalveti, can you point phh to the compiz gles stuff ? Apr 18 15:28:17 * ogra_ forgot where it was, some upstream git tree iirc Apr 18 15:28:19 (the sources, i'm still on maverick.) Apr 18 15:28:38 or Amaranth ^^^ Apr 18 15:29:23 http://git.compiz.org/~amaranth/mobilebling/ Apr 18 15:29:37 Still working on the plugins Apr 18 15:29:52 its just to test it on tegra graphics Apr 18 15:30:10 Ah, ok Apr 18 15:30:16 Amaranth: just tell me some features that are supposed to work :p Apr 18 15:30:28 drawing windows and frames :P Apr 18 15:30:34 so compile and install that then run `compiz --replace composite opengl move resize decor` to get a basic window manager running Apr 18 15:30:48 Although all the plugins that get compiled in there should work Apr 18 15:30:52 ok Apr 18 15:31:10 The plugins-main pack is still being worked on, I thought I had it done then discovered weird things were happening Apr 18 15:31:48 The plugin pack is still compiling against regular GL and when running if it tries to make a regular GL call that function just gets skipped Apr 18 15:31:59 Weirdest thing ever, I would have expected a symbol error Apr 18 15:32:29 blame the toolchain :P Apr 18 15:32:33 So I haven't published it yet, even though most of it actually works Apr 18 15:33:33 I've got the build system putting the proper defines in the pkg-config file now at least but the plugin pack seems to ignore that anyway Apr 18 16:42:11 Amaranth: would you mind adding a vendor string matching for image_pixmap extension ? nVidia doesn't declare it, even though they have it Apr 18 16:43:37 phh: ? Apr 18 16:44:03 nvidia doesn't report supporting the standard one but only their own version? Apr 18 16:44:24 no Apr 18 16:44:32 they don't report anything concerning this thing Apr 18 16:44:41 but they do support the extension Apr 18 16:44:46 *headdesk* Apr 18 16:44:54 :) Apr 18 16:45:40 phh: Do they at least properly report support for GL_EXT_texture_format_BGRA8888 or should I add a workaround for that too? Apr 18 16:46:01 I need to add one of those for the efika anyway, the current workaround only works if you compile on the efika with their headers Apr 18 16:47:41 they report GL_EXT_texture_format_BGRA8888 Apr 18 16:47:57 phh: What are the EGL_VENDOR and GL_RENDERER for that system? Apr 18 16:49:15 GL_VENDOR: NVIDIA Corporation Apr 18 16:49:15 GL_RENDERER: NVIDIA Tegra Apr 18 16:50:25 Amaranth: so, compiz doesn't crash, but no alt-tab (I guess I'm missing some missing files), and I've got double buffering problem (like on kwin) Apr 18 16:50:47 basically, when there are updates, only part of the buffer are updated and only on the current buffer, the second buffer doesn't get the update Apr 18 16:50:58 cd Apr 18 16:53:17 ndec: ~ Apr 18 16:53:30 rm -rf / Apr 18 16:55:14 ndec: rm: il est dangereux d'opérer récursivement sur «/» Apr 18 16:55:14 rm: utilisez --no-preserve-root pour inhiber cette mesure de sûreté Apr 18 16:59:57 phh: alt-tab doesn't work because you didn't load a plugin for it :) Apr 18 17:00:21 I'll have to see what was done for kwin as far as the other problem, I remember seeing something about it Apr 18 17:00:33 iirc it hasn't been fixed yet :s Apr 18 17:00:48 Hmm, I could have sworn jazh had a fix committed for it Apr 18 17:01:30 in screen.cpp, line 1268, doing mask=COMPOSITE_SCREEN_DAMAGE_ALL_MASK works Apr 18 17:01:32 but hum. Apr 18 17:02:04 Amaranth: loadign switcher is supposed to work ? Apr 18 17:02:23 compiz (core) - Error: Plugin 'compiztoolbox' not loaded. Apr 18 17:02:23 compiz (core) - Error: InitPlugin 'switcher' failed Apr 18 17:02:23 compiz (core) - Error: Couldn't activate plugin 'switcher' Apr 18 17:02:23 compiz (core) - Error: Plugin 'compiztoolbox' not loaded. Apr 18 17:02:43 ah I have to load it by hand before switcher ? Apr 18 17:03:06 Right Apr 18 17:03:09 ok. fine. Apr 18 17:03:37 interesting. Apr 18 17:05:02 That's not the normal switcher we use but it should work Apr 18 17:05:25 Although I think I may have broken the icon loading, I need to go back in and put a proper RGBA->BGRA conversion in Apr 18 17:05:39 what switcher should I use ? Apr 18 17:05:45 I agree it's kind of buggy :p Apr 18 17:05:52 The normal switcher is in plugins-main Apr 18 17:06:07 Which I've spent about a week fighting cmake over Apr 18 17:06:12 ok Apr 18 17:06:40 I basically have to somewhat drastically change the plugin build system in order to get plugins-main to pick up on the fact we're using GLES Apr 18 17:06:45 so, the bugs I have, two out of 3 windows are overbrightnessed or something like that Apr 18 17:07:03 (could be because of the hack I just added) Apr 18 17:07:12 Which hack? Apr 18 17:07:28 mask=blabla Apr 18 17:07:38 panda and efika don't have such a problem (although I suppose it's hard to tell with efika considering the other issues there) Apr 18 17:07:39 alt-tab has black windows, until windows are refreshed Apr 18 17:07:51 Weird... Apr 18 17:08:10 icons are black too Apr 18 17:08:21 all these problems could be because of my hack :D sec I test without it. Apr 18 17:08:45 The icons are expected Apr 18 17:12:54 ok, still same bugs without my hack Apr 18 17:13:07 Amaranth: would it sounds stupid if it does the sum of the two last frames ? Apr 18 17:13:31 I can't think of why that would happen Apr 18 17:13:55 That doesn't sound like something I'd do accidentally, anyway Apr 18 17:17:12 only app that isn't disturbed is the terminal which uses alpha channel ... Apr 18 17:52:04 Looks like do-release-update May be a bad idea for our images. Took 3 days (largely due to user prompts for config file overwrites) and X fails to come up afterwards. Looking into the issues now. Apr 18 17:52:21 Also, this will fail to get any x-loader & u-boot updates. Apr 18 17:52:42 (which is why I recommend that they be installed as packages). Apr 18 20:40:06 Hmmm. I had two pandas running do-release-upgrade over the weekend. My 8G failed for unknown reasons (looks like it may have had the pvr drivers for maverick installed prior to upgrade). The 16G SD upgraded fine. Apr 18 20:40:47 Although the partition label for the boot partition was deleted. **** ENDING LOGGING AT Tue Apr 19 02:59:57 2011