**** BEGIN LOGGING AT Wed Sep 15 02:59:57 2010 Sep 15 08:19:13 Morning sebjan Sep 15 08:22:09 Morning ndec Sep 15 08:23:20 Morning anyone else at TI :) Sep 15 08:24:01 good morning lag Sep 15 08:24:09 morning lag :) Sep 15 08:24:18 persia: looking at https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/637947 Sep 15 08:24:20 Launchpad bug 637947 in linux-ti-omap4 (Ubuntu Maverick) (and 1 other project) "no sound devices on current ES2.0 boards (affects: 1) (heat: 10)" [High,Confirmed] Sep 15 08:24:23 Has a meeting just finished? Sep 15 08:24:24 :) Sep 15 08:24:45 sebjan: http://launchpadlibrarian.net/55654467/dmesg.txt :( Sep 15 08:25:25 persia: I don't get how to translate our daemon.conf settings into this /usr/share/alsa/alsa.conf file. It doesn't sound to me we can even change this alsa.conf file either... Sep 15 08:25:55 cooloney: the last patches submitted into the ti-omap4 ubuntu tree are not in a package yet. There has been a start new release, but no new package version commit. Can we expect to have this commited before the freeze? Sep 15 08:28:35 lag: high-res timers flag appears to be deactivated into our kernel config, though I can't'understand why... this might be the problem? Sep 15 08:29:01 So, what do we need to do to change it? Sep 15 08:29:34 well, fdr editconfigs ;) Sep 15 08:29:50 On it's just a config option? Sep 15 08:29:51 berco, At a broad level, we don't want to change the configuration files unless we're absolutely sure that we're changing them in a way that is the best set of defaults for every potential platform on which Ubuntu is installed, which makes it tricky. I'm looking through your attachments to see if I can come up with any translation guidance. Sep 15 08:29:55 Cool Sep 15 08:30:23 What's the difference between ti-ubuntu- and for-buntu- Sep 15 08:30:51 for-ubuntu-<> is the patches I propose to go into the official ubuntu tree Sep 15 08:31:07 ti-ubuntu-<> is th tree I use to generate images for our internal use Sep 15 08:31:47 sebjan: it is just a start new release commit, its not realy about kernel freeze Sep 15 08:32:10 both branches are almost aligned (except for the changlog that I hack in the ti-ubuntu-<> branch) Sep 15 08:32:20 sebjan: today is Sep 15, tomorrow will be kernel freez Sep 15 08:32:31 persia: thanks Sep 15 08:32:36 cooloney: I was just wondering when we would get these patches into a daily image :) Sep 15 08:33:17 lag: what did run to get the error your sent me (with the high-res timers message)? Sep 15 08:33:59 Totem Sep 15 08:34:35 sebjan: aha, i need to ping tim, who will try to bump it to 903.12 and upload it Sep 15 08:34:46 sebjan: then we got them in daily image Sep 15 08:35:02 cooloney, Just be careful with the metas: we had a fail-to-upload recently on linux. Sep 15 08:36:23 persia: oh, what's issue of fail-to-upload? Sep 15 08:36:58 I'll let you know as soon as I have free RAM :) Sep 15 08:39:57 cooloney: it seems that we miss the high-res timers in our defconfig. lag reported an error indicating they are deactivated, and indeed they are deactivated from our defconfig. Sep 15 08:40:30 sebjan: OK, got it, thx lag Sep 15 08:40:49 k Sep 15 08:41:08 sebjan: are you going to enable it? Sep 15 08:41:10 Do you want me to test and submit to the Kernel Team Mailing List? Sep 15 08:41:35 berco: yes Sep 15 08:41:53 sebjan: 10.10 or 24.9? Sep 15 08:42:01 berco: both Sep 15 08:42:10 sebjan: cool, thx Sep 15 08:42:17 lag: yeah, please Sep 15 08:42:44 lag: so if you enabled that in kernel, the alsa oops will be gone, right? Sep 15 08:43:13 * lag shrugs Sep 15 08:43:17 I'll know when I test Sep 15 08:44:15 lag and sebjan, i think we need to enable that Sep 15 08:44:21 $ grep -r CONFIG_HIGH_RES_TIMERS debian.master/config/ Sep 15 08:44:21 debian.master/config/i386/config.common.i386:CONFIG_HIGH_RES_TIMERS=y Sep 15 08:44:21 debian.master/config/config.common.ports:CONFIG_HIGH_RES_TIMERS=y Sep 15 08:44:21 debian.master/config/armel/config.flavour.omap:CONFIG_HIGH_RES_TIMERS=y Sep 15 08:44:21 debian.master/config/armel/config.flavour.versatile:# CONFIG_HIGH_RES_TIMERS is not set Sep 15 08:44:37 debian.master/config/amd64/config.common.amd64:CONFIG_HIGH_RES_TIMERS=y Sep 15 08:45:01 to align with other configs. Sep 15 08:45:14 cooloney, lag, yes I think so, and I checked it was also active in our dev team's tree Sep 15 08:45:29 sebjan: ok, great. Sep 15 08:46:02 * cooloney is trying to enable dynamic ftrace on omap4 to debug the highmem issue. Sep 15 08:46:17 i got some patches from Mathieu Sep 15 08:47:12 cooloney: cool if this works! (dynamic ftrace) Sep 15 08:48:41 cooloney, I can't find it offhand: had something to do with differing ABI numbers Sep 15 08:50:13 I tried to use ftrace on one of the Arm boards before and it murdered the system Sep 15 08:52:10 persia: is this one: https://lists.ubuntu.com/archives/kernel-team/2010-September/012754.html Sep 15 08:52:23 lag: yes, me too, this was not supported upstream for arm. I don't know if it changed in the last months? Sep 15 08:52:26 lag: don't scare me. man Sep 15 08:52:29 berco, I believe that the translation of the shellscript is fairly easy: "amixer cset name='Earphone Driver Switch' 1" would become "CTL{name}='Earphone Driver Switch', CTL{value}=1". I'd recommend going for the "all mixer config", so that the user has everything when they want to start playing. You'd put the configuration in a separate file in /usr/share/alsa/init and then reference it from /usr/share/alsa/init/00main conditionally based on CA Sep 15 08:52:29 RDINFO Sep 15 08:52:49 cooloney, Precisely :) Sep 15 08:54:45 berco, For the default.pa static loading: what issues have you had with module-udev-detect? Sep 15 08:55:45 berco, Also, why do you want to disable module-console-kit in default.pa? We rely on this to support user switching (among other features). Sep 15 08:57:02 berco, lastly, the changes to daemon.conf are documented as bug #623242 Sep 15 08:57:03 Launchpad bug 623242 in pulseaudio (Ubuntu) "speex-float-1 provides poor performance on armel (affects: 1) (heat: 173)" [Low,Triaged] https://launchpad.net/bugs/623242 Sep 15 08:58:38 lag, Did you ever get anywhere with a board-specific driver for 637947? Sep 15 09:01:17 * persia continues to think that fiddling with configuration isn't the best way to address 637947 Sep 15 09:02:14 persia: I'm in mid swing Sep 15 09:02:27 Excellent! Sep 15 09:02:30 persia: Apparently sound works on the 8 layer board Sep 15 09:02:44 \o/ Sep 15 09:02:48 persia: Other people have stated that it works on the 6 layer too Sep 15 09:02:51 who cares about the rest :P Sep 15 09:03:02 sound "working" isn't the issue: it's whether the drivers are correctly reporting the wiremap to userspace. Sep 15 09:03:04 Did you see my latest post? Sep 15 09:03:23 the dmesg one? Sep 15 09:03:29 persia: I don't know anything about a kernel->user-space wiremap Sep 15 09:04:25 lag, well, it doesnt expose the HW to userspace at all ... (in the HW tab of the settings window) so something is still wrong Sep 15 09:04:43 lag, Hrm. Typically all this happens inside the kernel, and userspace just works. Let me find the right bit in the linux source docs. Sep 15 09:04:54 k Sep 15 09:06:03 Documentation/sound/alsa/soc/jack.txt has a description of how the jack description works for ASoC Sep 15 09:06:13 * ogra tests the netbook-launcher fixes ... Sep 15 09:06:32 But I think the stuff in machine.txt is the missing bit, from what I've heard (I'd be glad to be wrong) Sep 15 09:07:42 My understanding is that we have a platform driver for omap4, but no panda-specific driver. Sep 15 09:08:05 (all this is based on hearsay: I've not dug into the kernel code) Sep 15 09:08:45 sebjan: What's your knowledge on this? --^ Sep 15 09:08:55 lag, Anyway, for your dmesg, were you using the mangled config fragments berco posted? Sep 15 09:08:58 sebjan: Who do I need to speak to for more information? Sep 15 09:09:30 persia: I'm using TI's for-ubuntu kernel 'as is' Sep 15 09:10:13 I'm talking about userspace config Sep 15 09:11:02 There are ways to work around buggy kernels and buggy drivers in userspace, and we have a static solution for that, although it needs work to be available if we really can't fix it in the kernel. Sep 15 09:12:29 persia: thanks. For the default.pa I will check with our internal audio team as this is their recommandation I just followed. Also trying to understand if all that is needed Sep 15 09:12:54 I am using robclark's method - I'll send you the email Sep 15 09:13:02 lag: not much knowledge... I would say Liam (lrg) is the one. Though I am not sure if he is available this week? Sep 15 09:13:35 lrg is on holiday this week Sep 15 09:14:06 * XorA has the slimlogic panda on my desk anyway Sep 15 09:14:57 berco, OK. If the kernel drivers are clean, udev should be reporting the right stuff to pulseaudio through module-udev-detect, so we would set any values we need to workaround module loading defaults with udev rules (although it makes sense to me to just set the correct defaults in the kernel modules themselves, as long as we're at it). For module-console-kit, I've already spoken to folks more knowledgeable about audio in Ubuntu than I, and we Sep 15 09:14:57 really, really, really, really don't want to disable it. Sep 15 09:15:37 berco: Are you a TI bod? Sep 15 09:15:47 making a panda/fixing panda audio conf should take about 2 hours for a kernel hacker with schematic, its really not hard Sep 15 09:16:22 XorA, Does my understanding that we need a machine-specific driver match that estimate? Sep 15 09:16:35 cp omap4430sdp.c panda.c edit Sep 15 09:16:54 Id be surprised if you needed to change more than about 20 lines of code Sep 15 09:17:03 XorA: Where can we get said schematic? Sep 15 09:17:07 should just be the wiring map Sep 15 09:17:16 lag: bug prpplague or any of the other TI dudes Sep 15 09:17:49 berco, Would you be able to hunt down a wiring map, and attach it to 637947? Sep 15 09:18:04 in fact you would probably spend more time in wasted discussion here than actually coding :-) Sep 15 09:18:12 lag: yes, work with ndec and sebjan. same team Sep 15 09:19:45 persia: not sure if I can publish this info yet. Let me check on that. Sep 15 09:20:29 berco, OK. I'm fairly certain we can't upload a driver without the information (doesn't have to be a formal diagram, just knowing what codec pins attach to what ports, etc.) :) Sep 15 09:20:38 XorA: Do you mean: sound/soc/omap/sdp4430.c Sep 15 09:20:39 ? Sep 15 09:20:50 lag: yeah, I guessed at name without source in front of me Sep 15 09:21:04 Who is: Misael Lopez Cruz Sep 15 09:21:06 I think the other name is the Xloader driver for the same thing Sep 15 09:21:23 persia: understood. I will see what we can share. Agree it can be useful. Sep 15 09:21:36 lag: he is a TI contracter somewhere in the world Sep 15 09:22:14 sebjan: for the bus error, did you see my updates? Sep 15 09:22:22 It seems a bit of a waste to duplicate an 800 line file to amend just 20 lines Sep 15 09:23:22 lag, The alternative is making it somehow support both sorts of HW and export that sensibly to userspace. if you want an example of how painful that can get, look at the Intel HDA driver :) Sep 15 09:23:22 lag: no reason they cant be combined with the right if(machine_is) stuff Sep 15 09:23:50 lag: I was just going for quickest coding route Sep 15 09:24:50 np Sep 15 09:24:54 Well I'm happy to take it on Sep 15 09:25:02 I just need the correct documentation Sep 15 09:26:03 * persia updates the bug with bits from the conversation Sep 15 09:32:43 cooloney: no, which updates? Sep 15 09:35:25 sebjan: oh, not big change, just reproduce it on my es2.0 board Sep 15 09:35:30 but it is not bus error Sep 15 09:35:51 its Unhandled fault: imprecise external abort Sep 15 09:41:25 persia: any idea for the problem of H/W not listed in the Volume Applet? If I do a cat /proc/asound/cards I see the SDP4430 Sep 15 09:44:59 berco, Could you please paste 1) the output of `amixer` and 2) the additional entries in syslog.conf from `kill -HUP $(pidof pulseaudio)` ? Sep 15 09:45:37 I have a suspicion it's a continuation of the lack of a machine driver, but that's a guess. Sep 15 09:47:02 persia: for #1 see here: http://pastebin.ubuntu.com/494079/ Sep 15 09:59:27 persia: on my board I don't have syslog.conf. How is it supported to be generated? I see a syslog file in /var/log but nothing related to pulse when kill -HUP is done on PA Sep 15 10:00:02 My apologies. It's the extra stuff in /var/log/syslog I want to see. Sep 15 10:00:29 Please, if you ever have trouble following my directions, ask me if that's what I really meant :) Sep 15 10:00:42 persia: np :) Sep 15 10:00:52 Do you get anything on STDOUT from kill -HUP on pulse? Sep 15 10:01:10 persia: nothing on stdout or syslog Sep 15 10:01:26 because the pulseaudio manpage says it's supposed to dump the status report to STDOUT or syslog on receipt of SIGHUP :( Sep 15 10:02:03 Did you do anything special to the system to start pulse? Sep 15 10:03:40 persia: nothing special. it just starts at boot time Sep 15 10:03:55 session I believe rather than boot Sep 15 10:04:21 So just default. Is this with or without module-console-kit? Sep 15 10:05:56 just verified, I'm w/o module-console-kit Sep 15 10:06:24 while i'm waiting for the audio team to respond to me as of why to commented it I can try to re-enable it Sep 15 10:07:01 Would you? I'm not quite sure how all that works together, but I wonder if there's some race condition on login without that enabled that makes things different than expected. Sep 15 10:07:25 That said, it's a complete mystery to me why pulse isn't dumping a status report, and I wonder if it's even running properly. Sep 15 10:07:51 persia: pulse is difficult... Sep 15 10:08:36 Take care: it can get a bit more confusing when you start looking for sample-accurate mixing and routing :) pulse is a nice compromise. Sep 15 10:24:12 persia: uncommented the console-kit line in default.pas and got the following in syslog after kill command: http://pastebin.ubuntu.com/494097/ Sep 15 10:25:15 heh. I knew we wanted that for something :) Sep 15 10:25:26 :) Sep 15 10:26:19 Sorry, it took me some time to do the test. I had problems to make the test but I think it was b'cos of wifi which I removed for now. Sep 15 10:27:41 OK. Pulse is definitely seeing useful syncs. Do you still have the cardcoded default output to "mm" in your configuration? Sep 15 10:27:51 Err, sinks Sep 15 10:27:52 * ogra writes a bug "Lennart Poettering is spamming my logs" :P Sep 15 10:28:15 ogra, Why? that syslog stuff only appears on request (SIGHUP) Sep 15 10:28:16 persia: yes Sep 15 10:28:29 persia, yeah, i'm not serious :) Sep 15 10:28:44 Excellent! That means the problem is with indicator-sound, about which I know nothing :) Sep 15 10:29:02 ronoc does :) Sep 15 10:29:05 so do I Sep 15 10:29:10 catch him and corner him Sep 15 10:29:14 he wrote it Sep 15 10:30:02 what is indicator-sound supposed to do exactly? Sep 15 10:30:28 It's the name of the package that provides what is perceived as a volume applet. Sep 15 10:31:06 its the new UI tool, though it doesnt provide the sound dialog Sep 15 10:31:15 only the panel item Sep 15 10:31:46 Oh. Anyway, looking at the code, it seems to all be wildly autodetecting anyway, which ought just work, since there are sinks. Sep 15 10:31:48 ok, got it Sep 15 10:33:42 * persia hugs `lsw` from suckless-tools Sep 15 10:33:53 the app that pops up if you click the sound preferences item in the menu is still gnome-volume-control i think Sep 15 10:35:55 That's confusing. gnome-volume-control is clearly just querying pulse for sinks and sources, and we know pulse has them. Sep 15 10:36:20 berco, Could you try installing pavucontrol and seeing if you see more there? Sep 15 10:36:59 persia: yes, I can certainly do that. Maybe after lunach ;) getting hungry... Sep 15 10:37:42 persia: for the amixer settings, can't they be in a omap4430.conf file in /usr/share/alsa/cards folder? Sep 15 10:40:07 ndec, around ? Sep 15 10:40:49 berco, It would be better to have them in the kernel, in sound/soc/omap/panda.c or similar, but in the worst case, yes. Note that both files contain essentially the same information (although I don't know that anyone here knows how to convert the one to the other) Sep 15 10:41:39 Also, be careful about naming: the wiremap (and therefore which bits you want to set as defaults, etc.) is board-specific, not SoC-specific. Sep 15 10:41:44 persia: I agree with the kernel change but meanwhile I need to find a solution Sep 15 10:43:29 berco, Well, you can certainly look at dropping stuff in /usr/share/alsa to help indicate how to configure stuff. Depending on what bit you want to configure, drop it in a different place. As long as it's correctly autodetecting based on the hardware, it ought work. Sep 15 10:44:02 That said, I expect that regardless of what ends up being created kernel-side, anything you do now will have to change yet again once there is something kernelside. Sep 15 10:45:18 I'm also not convinced that the fails-to-show-up-in-the-volume-applet issue is at all related to the configuration. I could be wrong, but I'd like confirmation from another tool (pavucontrol) to ensure it's not a UI issue alone. Sep 15 12:21:28 hmpf Sep 15 12:21:41 why does the Favorites category not work anymore Sep 15 12:29:26 berco: Any luck on those documents? Sep 15 12:31:18 lag: sorry, no answer yet. Waiting for the Us to wake up Sep 15 12:55:54 lag, You might try to guess based on what data is being reported by the platform driver, and what configuration amixer needs (in the shell script attached to the bug). I suspect at least 50% of the data you need (and perhaps all of it) is encoded there somehow. Sep 15 13:04:29 * ogra goes mad trying to make favorites appear Sep 15 13:05:02 * persia looks at the build log for liblauncher, seeing absolutely no reason they shouldn't be there Sep 15 13:05:22 well, we dont have any .desktop file for them yet Sep 15 13:05:29 i just added one 2h ago Sep 15 13:05:40 and try to find out since then why it doesnt show up Sep 15 13:06:28 i dont really care about the normal favorites ... what i try to do is to add a .desktop file that contains an apturl entry for all the TI PPA packages Sep 15 13:06:57 so that if you click on it, it fires up software center, shows the EULA and installs them Sep 15 13:07:17 seems we have that feature now Sep 15 13:07:19 It's not driven by .desktop files directly: they come from the gconf key, and then do a lookup on the desktop DB to show them. Sep 15 13:07:34 the gconf key ? Sep 15 13:08:41 how would a gconf key affect .menu files Sep 15 13:08:45 or .desktop files Sep 15 13:09:06 persia, which gconf key would that be ? Sep 15 13:10:12 Unless I just read the code wrong after you commented on it again 40 minutes ago, liblauncher populates Favorites based on the /apps/netbook-launcher/favorites key Sep 15 13:11:04 And the default list is set in /var/lib/gconf/une.default/%gconf-tree.xml Sep 15 13:11:22 bah, sigh Sep 15 13:11:39 What have you been trying to do? Sep 15 13:11:46 which points to unity only Sep 15 13:12:02 i was using a very old netbook-launcher source Sep 15 13:12:27 Ugh, missed that. Good catch. Sep 15 13:12:31 which used a .menu and a .directry file to add a Favorites category Sep 15 13:12:51 seems thats in the older sessings package then Sep 15 13:12:55 * ogra goes digging Sep 15 13:12:58 That was all sorts of standards incompliant, and the interface I complained about enough to learn how it was done and get it sorted sensibly :) Sep 15 13:13:07 i want to get that apturl stuff in befpore freeze Sep 15 13:13:38 So, I think there's some latent bug somewhere, because liblauncher still hardcodes /apps/netbook-launcher/favorites in the source. Sep 15 13:13:50 Try sticking something there, and see if it works :) Sep 15 13:15:14 morning Sep 15 13:16:48 looking at http://launchpadlibrarian.net/55622759/add-favorites.diff (which was included in the last netbook-launcher-efl upload), I believe that the liblauncher location ought be compatible with the --add-favorite option. Sep 15 13:17:13 thats using ~/.local Sep 15 13:17:46 i need something i can add to our system defaults Sep 15 13:17:57 It's using local gconf, isn't it? Sep 15 13:18:06 Set a couple, then ship them as gconf-defaults. Sep 15 13:19:06 i'll try Sep 15 13:20:58 Good luck. Stick your findings in 613612, and I'll see if I can do anything based on that in case you don't make it by EOD. Sep 15 13:33:02 no go :/ Sep 15 13:39:37 persia, where exactly do you see the /apps/netbook-launcher/favorites stuff ? Sep 15 13:40:09 jayabharath, what were your probs with the image ? (you pinged yesterday) Sep 15 13:41:11 ogra: They were related to long boot times at first boot... rsalveti & ubot2 pointed me to the jasper fs issue on OMAP4... I guess its a known active problem .. Sep 15 13:41:24 no. its not Sep 15 13:41:33 it should just work Sep 15 13:41:40 resizing takes 2:30 for me Sep 15 13:41:49 jayabharath: I tested with yesterday's image and it worked fine, and very fast Sep 15 13:42:03 then another 20sec for setting the defaults, then it reboots into the config tool Sep 15 13:42:20 it's weird that from time to time someone complains about broken fs while resizing the image... Sep 15 13:42:36 ok. will try the latest daily and see if that makes a difference. Sep 15 13:42:38 hi there Sep 15 13:42:40 the overall install should be done within 10-12 min until it starts a usable enduser desktop Sep 15 13:43:04 are the arm kernels compiled for each devices, or that is not possible? Sep 15 13:43:07 * jayabharath think 10-12 mins is like eternity in the embedded world :) Sep 15 13:43:14 maybe a problem while giving dd and flushing the writes to the disk, don't know... Sep 15 13:43:15 oem-config actually takes the longest time here Sep 15 13:43:37 jayabharath, whats an embedded world ? Sep 15 13:43:38 :P Sep 15 13:43:42 :D Sep 15 13:43:44 we dont do embedded :) Sep 15 13:43:58 12min are incredibly fast for the ubuntu world :) Sep 15 13:44:11 that's for sure Sep 15 13:44:31 even 15-20 would be :) Sep 15 13:45:34 oh yeah.. we get annoyed if our board doesent booth within 30 sec :)... we had <30 sec boottimes with OMAP2... :P (ofcourse not under ubuntu) Sep 15 13:45:42 rlameiro, for each subarchitecture (we have omap3 kernels that should work with many omap3 based boards ... omap4 kernels etc etc) Sep 15 13:46:06 ogra: yes i know that, but there are some especifics that dont work Sep 15 13:46:20 jayabharath, heh, 30sec ... u-boot takes that long to only load the kernel from SD Sep 15 13:46:23 for instance there is a setup for the beagle board, but not for the IGEPv2 Sep 15 13:46:59 ogra, well, video and audio doesnt work Out the box on the Igepv2 Sep 15 13:47:09 orga - thats true.. we did some hacks in uboot & we had NAND flash on those boards.. now its all SD... Sep 15 13:47:23 rlameiro, might be that there is a bug open for that already, if not, open one Sep 15 13:47:49 jayabharath, yeah, SD is the painful part ... the panda would really fly if we had faster disk IO Sep 15 13:48:04 but SD somewhat restricts that Sep 15 13:48:10 I'm using one with an usb disk and it's a lot faster :-) Sep 15 13:48:15 where do i open it? on the ubuntu -arm kernel team launchpad? Sep 15 13:48:15 seems like a normal desktop Sep 15 13:48:16 yeah Sep 15 13:48:47 rlameiro, just on launchpad against the linux package, but i'm pretty suzre there are already bugs for it, so search first Sep 15 13:49:14 ok Sep 15 13:49:18 ogra: thanks Sep 15 14:11:54 robclark: Morning Sep 15 14:12:38 gm lag Sep 15 14:12:57 robclark: Do you know who wrote the userspace hacks to make sound work on the Panda? Sep 15 14:13:30 ahh, you mean the default.pa and that stuff.. Sep 15 14:13:42 I guess someone on audio team, but I'm not sure.. Sep 15 14:14:42 Who are this mysterious 'audio team'? Sep 15 14:15:09 you can probably only reach them by phone :) Sep 15 14:15:29 well, lgr might be best to ping.. although I think he is out of office this week.. Sep 15 14:15:48 I'm not sure if rest of them are on IRC much.. but I can harass them to get on IRC Sep 15 14:16:00 He is on holiday this week Sep 15 14:16:09 That would be good :) Sep 15 14:16:56 I don't really want to waste hours learning things that can be imparted in minutes Sep 15 14:17:13 Even email is better than nothing/phone Sep 15 14:17:24 But IRC would be best, as other people can interject Sep 15 14:17:37 yeah, agreed Sep 15 14:18:00 Cheers buddy - let me know how you get on Sep 15 14:18:56 * robclark just tries to find email thread where we were discussion audio/pulse issues.. Sep 15 14:23:02 lag: sadly I don't think anyone really knows (TI) what we need to do with the audio config... and we just have this dirty workaround for now. Sep 15 14:23:42 lag: persia, ogra and berco have started some discussion on 637947 Sep 15 14:23:43 ndec: If I know what the dirty workaround/hack is, I can transcribe it into the kernel Sep 15 14:23:44 wohoo, finally ! Sep 15 14:23:54 * ogra sees favorites on the launcher Sep 15 14:24:23 ogra: sorry man! i was stuck in meethings so far and my irc had crashed... so no clue what was discussed so far... someone told me you were looking for me. Sep 15 14:24:31 637947 is assigned to me :) Sep 15 14:24:41 lag: cool... so I will just wait for the fix ;-) Sep 15 14:24:46 ndec, i mailed you instead ... with the apport instructions Sep 15 14:24:57 lag did you receive the default.pa and alsa conf file? Sep 15 14:24:58 I have seen/read that, but most of that discussion is guesses :) Sep 15 14:25:00 ndec, seems apport has all you need Sep 15 14:25:14 ndec: I did, but I don't know what's changed Sep 15 14:25:43 lag: with our custom default.pa and alsa conf, we are able to have audio at boot (include the boot welcome sound). we are able to switch dynamically from HMDI and speakers while totem is playing... Sep 15 14:25:52 i think persia added some proper info how the amixer stuff needs to be handled Sep 15 14:26:05 * ogra wouldnt call that guesses Sep 15 14:26:35 ogra: lag: I think the *real* fix is to fix the driver. a nice workaround is to create a hard coded config in /usr/share/also/.../SDP4430, a dirty workaround is our .sh script that runs amixer config Sep 15 14:27:00 ndec, no Sep 15 14:27:06 ogra: thanks for apport stuf.. i will check this. how about the PPA armel for tiomap-dev? Sep 15 14:27:10 ndec, there are better ways in alsa to achieve what you need Sep 15 14:27:22 thats what persia elaborated on in the bug Sep 15 14:27:30 I'm happy to fix the driver Sep 15 14:27:37 heh Sep 15 14:27:39 ogra: but there are pulse config problems too. Sep 15 14:27:58 ndec, the two changes to default.pa are in now Sep 15 14:28:08 and it seems the support for the config dir doesnt work Sep 15 14:28:09 ogra: default.pa should figure out what the output/input audio are from /proc/asound, right? today we need to hard code the streams in default.pa Sep 15 14:28:38 well, there were two lines that needed changing in default.pa iirc Sep 15 14:28:42 these went in Sep 15 14:28:47 ogra: i think you are talking about daemon.conf changes. we also have changes in default.pa to create/list the output and input streams Sep 15 14:28:50 with todays upload Sep 15 14:29:04 no, i think i talk about default.pa Sep 15 14:29:08 * ogra checks Sep 15 14:29:46 bug 623242 Sep 15 14:29:47 Launchpad bug 623242 in pulseaudio (Ubuntu) "speex-float-1 provides poor performance on armel (affects: 1) (heat: 173)" [Low,Triaged] https://launchpad.net/bugs/623242 Sep 15 14:29:51 * ndec checks too Sep 15 14:30:07 gah, you are right Sep 15 14:30:22 +default-fragments = 8 Sep 15 14:30:22 +default-fragment-size-msec = 10 Sep 15 14:30:37 and +resample-method = speex-float-1 Sep 15 14:30:43 three lines actually Sep 15 14:30:44 ogra: yes, in daemon.conf.in Sep 15 14:30:54 yup Sep 15 14:30:58 sorry, my bad Sep 15 14:31:02 ogra: np Sep 15 14:37:45 Right, let's start from the beginning Sep 15 14:37:54 What is the SDP4430? Sep 15 14:38:17 other name of blaze? Sep 15 14:38:58 Thanks Sep 15 14:39:16 What AudioIC/CODECs are on the Panda/Blaze? Sep 15 14:40:14 lag: hrw: SDP4430 is actually another board OMAP4430 based (Software Development Platform) Sep 15 14:41:15 Okay s/Blaze/SDP4430 Sep 15 14:41:24 lag: SD4430 is the name of the audio chip in our audio driver. cat /proc/asound/cards Sep 15 14:42:22 Now that makes more sense Sep 15 14:42:51 So why am I porting sdp4430.c if the board I'm porting to has an sdp4430? Sep 15 14:44:35 ndec, btw, i found a very elegant way for the ppa package install ... including to show a general EULA for the PPA itself Sep 15 14:46:45 ogra: interesting. but I think we will have different EULA for different packages... Sep 15 14:47:23 well Sep 15 14:47:54 Do the SDP4430 and the Panda have different CODECS? I know the SDP4430 has a TWL6040 - what about the Panda? Sep 15 14:48:22 ogra: i have reused the sun-java source package to make my EULA/clickwrap package based on debconf Sep 15 14:48:36 lag: same CODECS. only the name is confusing... Sep 15 14:49:28 ndec: So they have the same audio IC and the same CODEC? Sep 15 14:49:34 So what needs porting? Sep 15 14:49:35 ogra: we can potentially present a message to mention that user is about to user a PPA that comes with restrictions... I can get an official text. anyway you can demo that to me? Sep 15 14:49:43 lag: yes Sep 15 14:50:18 lag: well it does not work on both ;-) if you boot a ubuntu image, you have no audio until you run our customer amixer.sh script and until you hack default.pa Sep 15 14:50:55 ndec, not yet ... but the plan is to have an icon in the favorites category that will trigger software-center using apturl its very elegant and can show a text as html Sep 15 14:51:09 ndec, https://wiki.ubuntu.com/SoftwareCenter#apturl Sep 15 14:51:31 ndec: :) Sep 15 14:51:35 i need to get your public PPA working first, and one package in there, then we can test it Sep 15 14:51:47 * ogra still waits for reply for the PPA Sep 15 14:51:52 I'm aware it doesn't work on both - I'm trying to find out why i.e. what's different Sep 15 14:52:26 ogra: once the public PPA is armel ;-) I can push a package there. Sep 15 14:52:38 yeah Sep 15 14:52:41 i'm waiting for IS Sep 15 14:53:00 ogra: that would be a GPL package (e.g. no debconf mess) but that should be fine for you, right? Sep 15 14:53:03 i suddenly have to use a new process ... that sadly takes longer :( Sep 15 14:53:13 anything is fine :) Sep 15 14:53:25 as long as a .deb comes out of it ;) Sep 15 14:53:41 ogra: is that really important ;-) Sep 15 14:54:46 ogra: the apturl is indeed really cool.. you are a lucky guy, the work is done already! Sep 15 14:54:56 lag, AFAIK audio is same on panda and sdp/blaze.. both have TWL6040, tho panda might be a newer revision (not 100% sure about that) Sep 15 14:55:07 ndec, yeah :) Sep 15 14:55:39 ndec, about bug 631362 ... we should try to join forces to get it fixed, seem luke would accept a patch if he could get one that doesnt crash pulse Sep 15 14:55:41 Launchpad bug 631362 in pulseaudio (Ubuntu Maverick) (and 1 other project) "Include several configuration files (affects: 1) (heat: 14)" [Medium,Triaged] https://launchpad.net/bugs/631362 Sep 15 14:56:35 ogra: welll. maybe.. that would give us an elegant solution to implement a dirty workaround and move our customization from default.pa into omap4.pa... Sep 15 14:56:46 right Sep 15 14:56:56 i dont see any other way here Sep 15 14:57:05 you cant modify default.pa Sep 15 14:57:11 ogra: from the log the commit I am interested in, is breaking pulse... that looks weird.. Sep 15 14:57:19 unless you ship your own copy of pulse Sep 15 14:57:23 ogra: well, I can modify it.. it's just uggly... Sep 15 14:57:35 not in a distro way Sep 15 14:57:48 the only possible solution would be to ship your own pulse Sep 15 14:57:52 ogra: how about divert? Sep 15 14:57:57 or to put up a HOWTO Sep 15 14:58:21 if you divert you will have nedless problems for people that upgrade Sep 15 14:58:32 diversion is never a solution Sep 15 14:58:38 *endless Sep 15 14:58:39 ogra: don't scream but right now, we have a meta package that makes a backup and override the file in preinst and restore in prerm ;-) Sep 15 14:58:50 * ogra screams Sep 15 14:58:52 :P Sep 15 14:58:55 lol Sep 15 14:59:08 lets try to get pulse fixed Sep 15 14:59:21 ogra: i am claiming we do everything right ;-)... no even claiming we do many things right! Sep 15 14:59:37 * ndec knew it would be a good motivation for ogra Sep 15 14:59:37 heh Sep 15 14:59:59 if i have solved the PPA stuff i'll take pulse and see what i can do Sep 15 15:00:07 iirc default.pa is actually a perl script Sep 15 15:00:55 it might even be possible to do the include from within default.pa Sep 15 15:01:13 and to just override the exiusting values from parsed files Sep 15 15:04:48 ogra: by the way, we might need stuff from universe in order to enable all omap4 features (such as obexftp). can you activate somehow unicers? Sep 15 15:04:56 s/unicers/universe/ Sep 15 15:05:42 hmm Sep 15 15:05:55 i can try to hack something into jasper Sep 15 15:06:17 jasper turns more and more into "hack around that prob" anyway Sep 15 15:06:31 oh, no, wait Sep 15 15:06:41 ndec, thats only for ppa stuff, right ? Sep 15 15:06:59 ogra: yes. this is for using bluetotth Sep 15 15:07:07 apturl ;) Sep 15 15:07:13 it will be able to take care Sep 15 15:07:25 ogra: too easy.. Sep 15 15:07:25 the best thing since sliced bread Sep 15 15:07:30 (and butter) Sep 15 15:08:33 ndec, do you have any TI logo i could just use for the desktop icon (else i'll make something up or find a panda picture or some such) Sep 15 15:11:57 * ogra takes a (really needed) break Sep 15 15:21:30 rsalveti: it seems that we are missing a patch into the x-loader embedded into the daily images. Where can I find the tree for this x-loader, to check all the patches are integrated into it? Sep 15 15:22:52 sebjan, the copyright file in /usr/share/docs/ usually has it Sep 15 15:23:03