**** BEGIN LOGGING AT Tue Sep 14 02:59:57 2010 Sep 14 07:23:27 morning Sep 14 08:42:46 ogra: does USB EHCI of ES2.0 work with our Ubuntu daily live image? Sep 14 08:43:15 yep Sep 14 08:43:57 kbd and mouse work fine on 6 and 8 layer for me Sep 14 08:44:15 the NIC too (usb NIC) Sep 14 08:45:27 nice Sep 14 08:46:33 ok, let me try again. Sep 14 08:46:35 thx, man Sep 14 08:46:53 do you have issues ? Sep 14 08:47:37 * ogra_ac didnt try todays image, yesterdays worked fine Sep 14 08:50:12 ogra_ac: i am trying Sep 11 daily image Sep 14 08:50:29 but usb EHCI doesn't work Sep 14 08:52:48 weird Sep 14 08:53:19 cooloney, 6 or 8 layer ? Sep 14 08:54:30 ogra_ac: 8 layers Sep 14 08:55:31 ogra_ac: i tried EHCI ports before with maverick rootfs from rootstock Sep 14 08:55:38 it works on my 8 layers Sep 14 08:56:22 well, all i can say is that it works flawless here Sep 14 08:56:23 ogra_ac: if you think yours is working fine, never mind. maybe it's my bad. Sep 14 08:56:31 ogra_ac: ok, got it. Sep 14 08:56:40 with teh dsailsy it should just work for you too Sep 14 08:56:46 *daily Sep 14 08:57:27 * ogra_ac recommends to zsync to the latest Sep 14 08:58:53 though there wasnt a kernel change between 11th and today afaik Sep 14 09:00:42 oh, wait Sep 14 09:00:54 the metapackage was screwed until teh 12th Sep 14 09:01:16 you might still have the old ES1.0 kernel on the image from the 11th Sep 14 09:01:21 oh, i'm just wanna zsync that Sep 14 09:01:29 ogra_ac: ok, i got it. Sep 14 09:01:39 ogra_ac: let me zsync it Sep 14 09:01:48 yeah, use todays Sep 14 09:02:03 intresting that it boots at all Sep 14 09:02:22 i think mine got stuck on teh 6 layer when i tried it with the old kernel Sep 14 09:02:31 during first boot Sep 14 09:05:02 ogra_ac: thanks for helping this. i'm zsyncing Sep 14 09:05:15 cool Sep 14 10:11:44 ogra: Is there any reason why the ES2.0 wouldn't boot with CONFIG_PM enabled? Sep 14 10:11:56 lag, ask ndec Sep 14 10:12:01 (if he returns) Sep 14 10:12:18 lag, all i was told is that the HW doesnt support it yet Sep 14 10:12:28 and wont for maverick Sep 14 10:12:42 k Sep 14 10:12:50 I've done some testing and it doesn't even boot Sep 14 10:12:56 No boot-up messages at all Sep 14 10:12:59 but ndec should be able to tell you if it gets in our way Sep 14 10:13:09 ah, so it does then :) Sep 14 10:13:09 I'll move on to something else for the time being Sep 14 10:13:20 Quite Sep 14 10:13:44 you could fix some userspace bugs if you are bored :P Sep 14 10:13:55 we have enough of them Sep 14 10:24:31 lag, how about ASoC stuff ? Sep 14 10:25:10 we still have no working sound and TI wants to use asound.conf,getting the driver improved to not need that would be an awesome fix Sep 14 10:29:26 ogra: I though it worked? Sep 14 10:29:33 ogra: Or was that ES1.0? Sep 14 10:29:37 HDMI might Sep 14 10:29:43 no it never worked properly Sep 14 10:30:19 for the normal system we should default to the earphone plug ... with an option to switch over to HDMI in the sound prefs Sep 14 10:30:35 prpplague was working on it Sep 14 10:30:42 I'm sure he submitted patches Sep 14 10:30:45 in ES1.0 i saw the HW at least but couldnt switch (and i think with that sound came out of HDMI) Sep 14 10:31:05 with ES2.0 i dont see a sound device at all atm Sep 14 10:31:22 lag, he is working on a very raw implementation that requires asound,conf Sep 14 10:31:33 Create a bug Sep 14 10:31:36 And assign me Sep 14 10:31:42 our audio team usually doesnt allow using that file Sep 14 10:32:16 (though we'd do it as a workaround from jasper if we cant get it fixed, but if there is any opportunity to get it fixed properly that would be better) Sep 14 10:52:48 ogra: Have you reported a bug yet? Sep 14 10:53:00 nop Sep 14 10:56:44 lag, bug 637947 Sep 14 10:56:45 Launchpad bug 637947 in linux-ti-omap4 (Ubuntu Maverick) (and 1 other project) "no sound devices on current ES2.0 boards (affects: 1) (heat: 8)" [High,New] https://launchpad.net/bugs/637947 Sep 14 11:01:29 Ta Sep 14 11:10:31 persia, i'm wondering if we couldnt just use a case statement for jasper cmdline parsing ... if anything is set we just dont touch it Sep 14 11:10:32 lag, I recall you and I discussing this about a month ago: it either needs proper quirking *OR* board-drivers. Sep 14 11:11:10 persia, talking about sound or PW ? Sep 14 11:11:13 err Sep 14 11:11:14 PM Sep 14 11:11:17 Sound. Sep 14 11:11:54 well, TI brought it up on friday and they say there is no time to fix it in driver on their side Sep 14 11:12:04 so asound.conf would be our last resort Sep 14 11:12:08 Assuming you're talking about bug #605831, the issue with a case statement is that we only want to set things if they *aren't* present. You'd have to track a lot of state. Sep 14 11:12:09 Launchpad bug 605831 in jasper-initramfs (Ubuntu Maverick) (and 1 other project) "Resolution should be taken from /proc/cmdline if provided (affects: 1) (heat: 73)" [Medium,Triaged] https://launchpad.net/bugs/605831 Sep 14 11:12:29 but i'D rather not use it if we can make it happen properly anyhow Sep 14 11:12:31 Those aren't the only choices (and asound.conf is a very poor choice). Sep 14 11:13:06 In addition to those two, we have 1) kernel quirking to hint various devices if they don't report properly (this is done *extensively* for Intel HDA, as an example). Sep 14 11:13:10 well, feel free to make suggestions for fallbacks on the bug above Sep 14 11:13:51 we can surely create an asound.conf from jasper as last resort ... any better solution is indeed preferred Sep 14 11:14:54 And 2) the extensible constructed ALSA configuration, which happens dynamically in /usr/lib/alsa-lib Sep 14 11:15:09 Please, please, don't create an asound.conf in jasper. Sep 14 11:15:19 If nothing else, that will likely break all USB audio. Sep 14 11:15:29 Potentially bluetooth as well. Sep 14 11:15:51 well, all suggestions are welcome Sep 14 11:16:08 if you have a better way we shoudl use that one Sep 14 11:16:20 Drop a conf fragment in /usr/share/alsa/ somewhere that only gets used for omap. Sep 14 11:16:57 We'll end up with broken support for omap boards other than those identified for testing, but it won't be more broken than an asound.conf, and will be much easier to fix. Sep 14 11:17:33 I *think* the best place is probably under /usr/share/alsa/init/omap or similar Sep 14 11:17:41 k, can you add a comment on the bug Sep 14 11:18:03 sure Sep 14 11:18:25 (make sure though thats its a fallback, preferred solution is still in kernel) Sep 14 11:19:25 persia: Yes we did Sep 14 11:19:41 prpplague: Are you here yet? Sep 14 11:23:34 ogra, please review that comment: I hope it's clear enough. Sep 14 11:23:34 lag: where is here? Sep 14 11:23:50 prpplague, In front of a device connected to IRC and viewing this channel :) Sep 14 11:23:51 i guess he meant there :) Sep 14 11:23:52 * prpplague is travelling today and be back in the office tomorrow Sep 14 11:23:57 hehe Sep 14 11:25:12 lag: whatcha working on now? Sep 14 11:26:22 prpplague, i just created bug 637947 for him Sep 14 11:26:24 Launchpad bug 637947 in linux-ti-omap4 (Ubuntu Maverick) (and 1 other project) "no sound devices on current ES2.0 boards (affects: 1) (heat: 8)" [High,Confirmed] https://launchpad.net/bugs/637947 Sep 14 11:26:35 (which is why you got pinged i assume) Sep 14 11:27:19 prpplague: Hi David Sep 14 11:27:34 We were speaking about audio a few weeks ago Sep 14 11:27:54 I believe you were going to release some patches to make it work on the ES1.0 Sep 14 11:28:16 A board driver? Sep 14 11:33:16 lag: yea there was a handoff, the announcement went on the pandaboard list Sep 14 11:33:37 lag: you can get the latest sources at gitorious/pandaboard Sep 14 11:33:46 lag: the L24.9 branches Sep 14 11:37:15 prpplague: And they will make sound work on the ES2.0? Sep 14 11:37:28 I believe we have the majority of L24.9 Sep 14 11:54:55 prpplague: When you say the pandabaord list, do you mean on IRC, or is there a mailing list? Sep 14 12:01:22 lag: the last audio patches I pushed on my tree (for-ubuntu-2.6.35) are post L24.9, and are not in prpplague's tree Sep 14 12:02:21 sebjan: I have just fetched from your tree, which is building now Sep 14 12:02:23 :) Sep 14 12:02:36 lag: ok, cool :) Sep 14 12:02:49 :) Sep 14 12:06:15 cooloney: ping Sep 14 12:22:09 lag: there is a mailing list Sep 14 12:22:43 I've has a quick Google for it, but to no avail Sep 14 12:22:52 Would you be so kind as to link me to it please? Sep 14 12:26:33 lag: email: pandaboard@googlegroups.com Sep 14 12:26:47 lag: Link: http://groups.google.com/group/pandaboard Sep 14 12:28:03 ogra: i just subscribed you to 587632. any chance this can be fixed for 10.10? the problem is that when gst-ugly is installed it will superseed gst-ffmpeg which works fine on arm. Sep 14 12:28:27 bug 587632 Sep 14 12:28:31 Launchpad bug 587632 in libmad (Ubuntu) "Sound very distorted on armel (affects: 2) (heat: 18)" [Undecided,Confirmed] https://launchpad.net/bugs/587632 Sep 14 12:28:48 berco: I did find that Sep 14 12:28:57 berco: Is it just a matter of joining that group? Sep 14 12:30:25 lag: I think so Sep 14 12:30:28 ndec, lets ask linaro if they want to work on a fix, else we can indeed build with -marm Sep 14 12:30:45 lag: honnestly I haven't tried yet Sep 14 12:30:47 Given the proximity to release, let's just use -marm Sep 14 12:30:59 I can't join Sep 14 12:31:08 You need to have a Google account Sep 14 12:32:28 lag: yes you always need a google account to subscribe google groups I believe. Sep 14 12:32:30 lag: I believe so. I just tried with my google account and I'm now a member of the group Sep 14 12:33:05 lag: you can create a new google account with an existing email address. in that case it would give you a google account without gmail. Sep 14 12:56:39 robclark:gm i saw some talk related to EDID ? Sep 14 12:56:56 hi mythripk.. Sep 14 12:57:28 yeah.. but mpoitier doesn't seem to be around atm.. Sep 14 12:57:52 hmm , I should be able to get the code with seperate edid.c by this weekend Sep 14 12:57:58 same thing we were talking about the other day, about how to support EDID for DVI/other drivers.. Sep 14 12:58:04 ok, very cool Sep 14 12:58:42 ya with your patch for DVI as well , i have the old regular code for EEDID Sep 14 12:58:58 ahh, hi mpoirier Sep 14 12:59:05 speaking of the devil :) Sep 14 12:59:09 indeed Sep 14 12:59:16 :) Sep 14 12:59:29 finally mpoirier and mythripk in same channel at same time ;-) Sep 14 12:59:57 we should remove europe ... that gets the timezones closer together :) Sep 14 13:00:10 heheh, I'm not sure if it works that way.. Sep 14 13:00:21 ogra: but I like Europe... Sep 14 13:00:25 great for vacation. Sep 14 13:00:27 mpoirier: mythripk had mentioned: "hmm , I should be able to get the code with seperate edid.c by this weekend" Sep 14 13:00:28 lol Sep 14 13:00:52 robclark: cool - on what processor ? Sep 14 13:00:54 mythripk: hm, also interested on that :-) Sep 14 13:01:27 robclark: rsalveti is heading that project now. Sep 14 13:01:30 hmm k then let me send the code for review once im done Sep 14 13:01:42 mythripk: cool Sep 14 13:01:57 would you like some heads up ? Sep 14 13:02:21 mythripk: yes please. Sep 14 13:02:30 mythripk: what you're planing to do for improving the edid code on omap4? Sep 14 13:02:38 the edid parsing Sep 14 13:02:39 mpoirier: I guess this would be on omap4 kernel, but we should start trying to apply some of the patches on both to align omap3 and omap4 code.. Sep 14 13:02:47 rsalveti: ^^^^ Sep 14 13:03:07 robclark: yep, that's what I was looking for Sep 14 13:03:18 trying to integrate at least the parsing code Sep 14 13:03:25 create a seperate edid.c file which will handle all the edid and eedid code Sep 14 13:03:26 yup Sep 14 13:03:26 rsalveti: I can assist if need be. Sep 14 13:03:26 and trying to use the common kernel api for it Sep 14 13:03:34 mpoirier: cool Sep 14 13:04:03 hmm ive seen some changes on omap3 done by srinivas i could send out those patches... Sep 14 13:04:14 rsalveti: just keep in mind, in some use cases you might have hdmi display, but no framebuffer, only v4l2 device.. Sep 14 13:04:42 yup , it is independent of v4l2 or fb and tied to hdmi.c Sep 14 13:05:58 rsalverti, rob and mpoirier do you have any specific suggestions... Sep 14 13:06:07 robclark: true, but it shouldn't be tied with fb Sep 14 13:07:33 mythripk: just one... ideally somehow the edid parsing code could return a list/array of supported timings. Because eventually we want omapfb to populate modedb table with all supported modes Sep 14 13:08:34 mythripk: you are miles ahead of me on the topic - I'll let you run with it. Sep 14 13:09:25 robclark:hmm , i could add that but problem is there could be duplicates and problem with number of entries.. Sep 14 13:09:57 becaue the 8 timing blocks timing data and the SVD block preferred timing data will overlap Sep 14 13:10:16 and for DVI the standard timing data also should be populated... Sep 14 13:13:58 mythripk: probably most important is the # of distinct resolutions.. if you have multiple supported timings that are same resolution, maybe it doesn't matter so much which you pick Sep 14 13:14:20 the use case is just letting userspace pick from available supported resolutions Sep 14 13:15:48 hmm ok then let me create a list and if the timing is valid , i shall populate all the values onto it... Sep 14 13:16:36 mythripk: I guess it shouldn't hurt, tho, if you populate the list with multiple timings that are same resolution.. I guess either way would work Sep 14 13:18:39 robclark: yup not parsing for a duplicate , i shall populate if valid ... only thing i need to do is convert Short video descriptor to resolution string.. Sep 14 13:19:31 mythripk: I think it is ok to be a list/array/iterator of struct omap_video_timings Sep 14 13:22:03 ya but SVD info will only give CEA code so need to populate it into a timing Sep 14 13:22:34 k then time to head for my bus , i shall send the edid.c by this friday or so... good day Sep 14 13:22:52 mythripk: I think you just need to loop thru the table of all timings, and see if it matches anything that we support? Sep 14 13:23:00 ok, c-ya later Sep 14 13:24:51 yes ... all_timings[code] will do ... Sep 14 13:26:25 lag, ogra@panda:~$ dmesg|grep -i under Sep 14 13:26:26 [ 424.812652] omapdss DISPC error: GFX_FIFO_UNDERFLOW, disabling GFX Sep 14 13:26:26 [73202.197601] omapdss DISPC error: GFX_FIFO_UNDERFLOW, disabling GFX Sep 14 13:26:34 got it twice since yesterday it seems Sep 14 13:26:57 That's the one! Sep 14 13:32:14 ogra: Which bootloaders are we using? Sep 14 13:32:27 ogra: I fixed this before by using the correct ones Sep 14 13:32:32 lag, most recent linaro package Sep 14 13:32:57 Do you have back log on here? Sep 14 13:33:10 this channel you mean ? Sep 14 13:33:12 sure Sep 14 13:33:41 robclark> lag: do you have x-loader w/ DDR timings tweaks and 1gb support? Sep 14 13:33:41 without that, I don't think DSS gets enough memory bandwidth to feed a 1920x1080 hdmi monitor.. Sep 14 13:33:55 yes, we have that Sep 14 13:34:08 x-loader is ours btw ... only u-boot comes from linaro Sep 14 13:40:33 lag, http://irclogs.ubuntu.com/2010/09/14/#ubuntu-arm.html is also available, in case you ever need backscroll and nobody here has it. Sep 14 13:43:03 persia: Bookmarked, thanks Sep 14 13:43:35 lag, that's just today's: you might want to bookmark something further up in the hierarchy :) Sep 14 13:43:55 persia: I did ;) Sep 14 13:44:02 * lag is not just a pretty face :) Sep 14 13:44:45 Is there a searchable version? Sep 14 13:47:06 I don't know of one currently maintained, other than using site: restrictions in popular search engines. Sep 14 14:06:14 mpoirier_: http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/gfxsdk/3_01_00_07//exports/OMAP35x_Graphics_SDK_setuplinux_3_01_00_07.bin Sep 14 14:06:24 this is the original SDK "tarball" Sep 14 14:06:42 heh Sep 14 14:06:44 rsalveti: let me take a look. Sep 14 14:06:45 with the latest version you can just wget it Sep 14 14:06:52 but it's huge, let me send you just the kernel source Sep 14 14:06:57 500MB Sep 14 14:07:15 this is just in case you want to check where I got them Sep 14 14:07:35 rsalveti: is this about the GPled ? Sep 14 14:07:47 mpoirier: https://code.edge.launchpad.net/~rsalveti/+junk/omap3-sgx Sep 14 14:07:58 mpoirier: only the kernel part is gpl Sep 14 14:08:11 mpoirier_: this is a package I created using m-a to test Sep 14 14:08:48 using the same kernel sources, you can check my branch and see the GFX_Linux_KM directory Sep 14 14:09:13 this is the directory they put the source of their kernel modules Sep 14 14:09:48 what exactly are we doing ? Sep 14 14:10:16 mpoirier_: the idea of getting this source is to support the powervr sgx acceleration Sep 14 14:10:20 mpoirier_, adding a 3D driver Sep 14 14:10:24 to have opengles support Sep 14 14:10:53 ok, nice to know. Sep 14 14:13:26 * persia idly wonders if the GPL module is extensible to also support other powervr sgx hardware Sep 14 14:14:54 \ Sep 14 14:16:11 for omap 4 it's probably a different code/module Sep 14 14:16:22 but not sure, never saw it Sep 14 14:19:02 mpoirier_: http://paste.ubuntu.com/493659/ Sep 14 14:19:07 the build log, in case you need it Sep 14 14:19:15 to check how I'm currently building Sep 14 14:21:09 mpoirier_: rcn-ee also did some work on it, when integrating the sgx into his kernel, but adding at staging Sep 14 14:21:12 http://bazaar.launchpad.net/~beagleboard-kernel/%2Bjunk/2.6.36-devel/files/head:/patches/sgx/ Sep 14 14:21:38 at http://bazaar.launchpad.net/~beagleboard-kernel/%2Bjunk/2.6.36-devel/annotate/head:/patch.sh#L126 you can see what patches are applied Sep 14 14:22:04 rsalveti: this is all very interesting. Sep 14 14:22:23 I'll start roaming through this stuff... Sep 14 14:23:54 mpoirier_: cool Sep 14 14:23:56 cool Sep 14 14:23:59 remember that this is just for omap 3 Sep 14 14:24:13 indeed Sep 14 14:39:54 -etoomanyogras Sep 14 14:40:08 heh Sep 14 14:40:09 still missing ogra_n900 :-) Sep 14 14:55:45 ogra/rsalveti... the boot time for the 10.10 daily images is about 10 mins on ES2.0 panda's is that what you are seeing on your end? Sep 14 14:56:18 jayabharath: for first boot (installer)? Sep 14 14:56:22 or any other normal boot? Sep 14 14:56:26 First boot Sep 14 14:56:35 rsalveti: on first boot Sep 14 14:56:43 * jayabharath thinks that it's trying to format SD card or something Sep 14 14:56:45 take a few to get up and running Sep 14 14:57:02 hm, the very first boot is the sd resizing Sep 14 14:57:08 it shouldn't take that long Sep 14 14:57:18 but it depends on your sd card size, I believe Sep 14 14:57:27 and the sd card speed Sep 14 14:57:28 4GB is what we are using... Sep 14 14:57:41 4gb class 4 should take 2, 3 minutes Sep 14 14:57:47 at least this is what I get here Sep 14 14:58:02 GrueMaster: did you tested latest omap 4 image already? Sep 14 14:58:15 rsalveti: do you get errors sometime when booting? Sep 14 14:59:05 orbarron|nb: no errors, just resize the sd card, reboot, oem-installer (around 10 min) and then the usual netbook interface Sep 14 14:59:17 Just downloading todays. Yesterdays worked fine. Sep 14 14:59:18 the only messages I'm getting is regarding sound Sep 14 14:59:30 hmm Sep 14 14:59:51 there's a bug that sometimes the first boot identifies a broken partition Sep 14 15:00:01 broken filesystem Sep 14 15:00:02 ok Sep 14 15:01:19 * orbarron|nb downloads today's image Sep 14 15:01:26 rsalveti: can you point me to the bug# Sep 14 15:01:37 looking for it Sep 14 15:03:47 hm, it sees that's a different issue, and with a quite old image: bug 613591 Sep 14 15:03:48 Launchpad bug 613591 in jasper-initramfs (Ubuntu) "Jasper sometimes fails to resize root partition on omap4 (affects: 1) (heat: 104)" [Undecided,New] https://launchpad.net/bugs/613591 Sep 14 15:04:07 * rsalveti also downloading latest image to test Sep 14 15:04:35 thanks ubot2/ rsalveti Sep 14 15:08:50 argh, getting a lot of smsc95xx 1-2.1:1.0: usb0: kevent 2 may have been dropped at my xM Sep 14 15:24:12 sebjan: ASoC Sep 14 15:25:56 lag: ASoC ? Sep 14 15:26:15 sebjan: ASoC Sep 14 15:26:17 :) Sep 14 15:26:38 "asoc: no valid backend routes for PCM: SDP4430 Media" Sep 14 15:27:36 lag: did you try to use the config files I sent earlier today? Sep 14 15:28:04 Nope Sep 14 15:28:07 Where do they need to go? Sep 14 15:28:36 Oh, I've just seen what you've written Sep 14 15:28:39 Give me a second Sep 14 15:31:36 sebjan: Still nothing Sep 14 15:31:43 But I am using the 6 layer board? Sep 14 15:32:47 lag: I did not test myself with 6 layers but I have heard of issues with it. I tested on 8 layers. Sep 14 15:33:34 Who knows more about these issues? Sep 14 15:37:49 orbarron|nb: resize went very fast, one minute or so Sep 14 15:37:53 4gb class 4 Sep 14 15:40:03 sebjan: How did you test? Application/file type? Sep 14 15:40:22 lag: do you have access to a 8 layers? Sep 14 15:40:34 sebjan: Unfortunately not Sep 14 15:42:00 lag: you may ask on #pandaboard if some hw differences could explain a difference... Sep 14 15:42:47 sebjan: Done Sep 14 15:43:09 ogra: Can you make me a member of Ubuntu armel porters? Sep 14 15:44:02 lag, done Sep 14 15:44:08 ta Sep 14 15:57:02 E: genext2fs: not enough memory for filesystem ??? Sep 14 15:57:57 how much memory does this thing actually need? Sep 14 15:58:20 Neko, exactly the size of the desired filesystem size Sep 14 15:58:27 like real memory? Sep 14 15:58:30 yes Sep 14 15:58:33 holy crap Sep 14 15:58:37 well, you can add swap Sep 14 15:58:43 but that will be very very slow Sep 14 15:59:04 i'D recommend not using the --no-root option unless you need to Sep 14 15:59:20 ouch, that's crazy! Sep 14 15:59:24 then it will just use a loop mounted image Sep 14 15:59:34 mopdenacker, it is Sep 14 15:59:44 but thats the way genext2fs is designed Sep 14 15:59:48 my concern is that my /tmp is tiny Sep 14 15:59:58 it allocates the whole image size in ram before copying contents Sep 14 16:00:51 I thought rootstock respected TMPDIR but it always makes the image in /tmp Sep 14 16:01:30 Could they do an mmap() on the output file instead of allocating RAM? This should be equivalent... Sep 14 16:02:15 I'm just gonna hack this so BUILDDIR has a template... Sep 14 16:02:34 mopdenacker, well, that no-root "feature" was only added for corner cases where people build on a big server but have no root access Sep 14 16:02:40 Thought they may not know how big the output is gonna be... Sep 14 16:02:59 the size is a parameter genext2fs expects Sep 14 16:03:02 so it knows Sep 14 16:03:30 its just very old software written at a time where you wouldnt have had the idea to roll gigabyte big images Sep 14 16:04:09 Neko, file a bug, that it should use TMPDIR Sep 14 16:04:18 I thought this was already a bug Sep 14 16:04:22 532342 Sep 14 16:04:29 bug 532342# Sep 14 16:04:30 Launchpad bug 532342 in rootstock (Ubuntu) "rootstock-gtk does not allow to specify the target rootfs tarball file path (affects: 1) (heat: 19)" [Medium,Fix released] https://launchpad.net/bugs/532342 Sep 14 16:04:39 rootstock-gtk Sep 14 16:04:40 hmm Sep 14 16:04:56 my mistake but it's the same idea :D Sep 14 16:05:12 well, rootstock-gtk surely respects it Sep 14 16:05:24 the gui doesn't have enough options for me :/ Sep 14 16:05:33 for a start, it doesn't have maverick in the distros list :D Sep 14 16:05:33 else it wouldnt be fix-released Sep 14 16:05:47 but please file a bug for the cmd Sep 14 16:05:59 or reopen it :-) Sep 14 16:06:08 well, its fixed in the gui Sep 14 16:06:10 okay it will have to wait until I have actually done what I wanted to do with it but I have a post-it note reminding me now Sep 14 16:06:22 i dont get how i fixed it there without fixing it in the cli though Sep 14 16:08:18 * rsalveti lunch Sep 14 16:09:18 just need to change it to BUILDDIR=$(mktemp -d ${TMPDIR}/tmp.XXXXXXXX) or so right? Sep 14 16:09:20 that's working here Sep 14 16:09:38 erm, no Sep 14 16:09:53 mktemp will use TPMDIR by default Sep 14 16:10:19 maybe mktemp doesn't do local shell variables? Sep 14 16:10:27 I have to really export it or something? Sep 14 16:10:31 ah, there is the bug :P Sep 14 16:10:51 -t interpret TEMPLATE as a single file name component, relative to a directory: $TMPDIR, if set Sep 14 16:10:58 man mktemp Sep 14 16:11:01 just add -t Sep 14 16:11:07 mktemp -d -t ? Sep 14 16:11:20 yep Sep 14 16:11:38 do I still need to file a bug? :D Sep 14 16:11:39 and file a bug so we dont forget to add it to the code Sep 14 16:11:52 doing it now Sep 14 16:13:16 Neko: bug and/or a patch Sep 14 16:13:24 :-) Sep 14 16:13:54 relly strange ... looking at #linaro i really dont get how they have so many probs with omap3 Sep 14 16:14:56 uhhhh mktemp -d -t did not fix it Sep 14 16:14:59 it still made one in /tmp Sep 14 16:15:19 how did you set TMPDIR ? Sep 14 16:15:29 TMPDIR="/build/rootstock/tmp" Sep 14 16:15:56 then I run rootstock in the next line of the script Sep 14 16:16:21 should I be putting it in front of the script? I'm not entirely sure how or why bash mangles these things.. Sep 14 16:16:58 yep that fixed it Sep 14 16:17:06 TMPDIR=whereveer rootstock blah blah blah works Sep 14 16:17:11 TMPDIR=whatplace Sep 14 16:17:16 rootstock blah blah does not :D Sep 14 16:17:33 it's funny because when you bash -ex it, the output is identical Sep 14 16:18:37 bug 638190# Sep 14 16:18:41 Launchpad bug 638190 in rootstock (Ubuntu) "rootstock cmdline does not respect TMPDIR (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/638190 Sep 14 16:20:07 ogra, what are all these omap3 probs they have? D Sep 14 16:20:58 oh shit I have a ubquity bug as well Sep 14 16:20:58 Neko, you ned to export TMPDIR if you dont use it on the same line Sep 14 16:21:11 simple shell rule ;) Sep 14 16:21:14 guys did you notice that when oem-config creates a user it does not put it in the audio group? Sep 14 16:21:47 from the installer (or rootstock even it seems) you get audio, but as soon as you oem-config in a new user, that guy gets no boot sound Sep 14 16:22:00 there is no audio group in ubuntu Sep 14 16:22:09 since ... like ... three releases or so Sep 14 16:22:25 its all handled via policykit and pulse Sep 14 16:23:09 I just had to fix it in my first maverick rootfs.. users and groups advanced, add to the audio group? Sep 14 16:24:36 if rootstock added users to the sudio group in lucid thats definitely a bug Sep 14 16:24:49 (none we'll fix now though :) ) Sep 14 16:25:00 *audio group Sep 14 16:26:41 maybe I am just overthinking it and mistaking the karmic demo image with the one I am working on :/ Sep 14 16:26:55 I did think it weird that the sd card from fsl did not have a user with rights to audio Sep 14 16:27:16 but my first xfce on maverick.. no audio.. ohhh I remember.. Sep 14 16:27:23 somehow it had no audio devices Sep 14 16:27:30 same symptom different problem Sep 14 16:27:48 ah, you have one of the babbage boards with the preinstalled SD card images they hacked together ? Sep 14 16:27:54 oh hell no Sep 14 16:28:17 we have the efika mx smarttop and smartbook with a hacked, improved version of their preinstalled sd card image :D Sep 14 16:28:26 heh Sep 14 16:28:27 like no FSL proxy settings Sep 14 16:28:50 I already complained profusely about how braindead it was. we also know the name of the system and the user who set it up. they did not clean it properly before shipping it. Sep 14 16:29:08 oh, i just realize who you are :) Sep 14 16:29:12 hi matt :) Sep 14 16:29:15 hi :) Sep 14 16:31:39 ogra, just to let you know, I received the bug assignment in my email this morning and will look into it today. May be a little difficult to test as I don't have the arm hardware yet, but I will read over the code and see if I can find the issue and get a fix for you Sep 14 16:32:03 ogra, btw rootstock package cache confuses the crap out of me. how come it does not manage to get new packages that aren't in the cache? Sep 14 16:32:21 if I save a cache and then try and re-use it again, but I've added a new package to the seed list, it will bomb Sep 14 16:32:36 why would the cache override mirror usage? shouldn't it pull updates since it's using apt anyway? Sep 14 16:32:53 devilhorns, you can test on x86 Sep 14 16:33:04 ogra, ok so it's not specific to arm then ? Sep 14 16:33:19 (haven't read through the whole email yet) Sep 14 16:33:26 devilhorns, the code we need ported already exists in the netbook-launcher package and needs to go into the netbook-launcher-efl code Sep 14 16:33:37 right, its not arm specific Sep 14 16:33:41 ah ok :) Sep 14 16:33:51 actually we very rarely have arm specific stuff to do in the UI Sep 14 16:34:03 if we get GL libs that might change Sep 14 16:34:14 but up to now you can do all work on x86 Sep 14 16:34:16 I see ... ok fair enough :) I'll look into this today Sep 14 16:34:36 ogra, we're going to look into that once we have a stable maverick for everyone internally to mess with Sep 14 16:34:40 Neko, i should have called it --flight-mode ... Sep 14 16:35:09 we have the GL lib source so we can build mx51 gl libs any way you like.. problem is mx51 is dropped and it needs our kernel (or at least, fsl 10.07) Sep 14 16:35:15 its for moments where you *really* dont have network access ... like on a plane, but want to do multiple image rebuilds Sep 14 16:35:29 (the package cache thing) Sep 14 16:35:38 any chance for maverick+0.5 to have a package cache with networking? Sep 14 16:35:51 if you have network access, just use a package proxy on localhost Sep 14 16:36:04 * ogra recommends approx Sep 14 16:36:07 that means I have to mirror 10GB Sep 14 16:36:11 no Sep 14 16:36:22 approx will onyl pull what you actually use Sep 14 16:36:29 oh cool Sep 14 16:36:54 ogra@osiris:~/Devel/packages$ du -hcs /var/cache/approx/* Sep 14 16:36:54 72M /var/cache/approx/ubuntu Sep 14 16:36:54 2,0G /var/cache/approx/ubuntu-ports Sep 14 16:37:03 i'm using that for image builds since jaunty Sep 14 16:37:25 quick setup instructions? Sep 14 16:37:34 for a single image build for maverick it shouldnt be above 5-600M Sep 14 16:37:41 apt-get install approx Sep 14 16:37:45 yes that's done of course Sep 14 16:37:53 but do I just pass rootstock this as a mirror? Sep 14 16:37:59 or do I need to really configure it Sep 14 16:38:18 edit /etc/approx/approx.conf Sep 14 16:38:28 what I have here is a really decent internet connection but ports.ubuntu.com is like 60k/s slowpoo. Sep 14 16:38:31 add: ubuntu-ports http://ports.ubuntu.com/ubuntu-ports Sep 14 16:38:33 at the top Sep 14 16:38:36 thats it Sep 14 16:38:36 cool Sep 14 16:38:43 http://www.engadget.com/2010/09/05/efika-mx-smartbook-now-on-sale-for-an-exceedingly-unattractive-p/ Sep 14 16:38:48 yes :) Sep 14 16:38:58 and for rootstock use the --mirror option Sep 14 16:39:07 $350. Sep 14 16:39:12 make sure to use your *external* ip there Sep 14 16:39:18 not localhost Sep 14 16:39:42 external == ifconfig eth0 ip address is okay? Sep 14 16:39:56 I'm more curious what 3D support there is. Sep 14 16:40:17 Any texture_from_pixmap support?? Sep 14 16:40:19 DanaG, none yet, but we're testing.. we found some serious performance snafu with the EGL X driver Sep 14 16:40:34 no texture_from_pixmap is a braindead desktop opengl thing Sep 14 16:40:47 you get that for free with embedded gl with EGLSurfaces Sep 14 16:41:13 or something like that Sep 14 16:41:38 to be honest I am from the braindead desktop gl world so embedded gl scares the crap out of me. I barely got used to not having a transformation pipeline. Sep 14 16:42:39 For me, I really want Compiz. Anything that can't do compiz is a non-starter. Sep 14 16:42:47 compiz is evil Sep 14 16:42:48 Well, kwin is acceptable, as well. Sep 14 16:43:04 they do not care unless it is amd64 with the nouveau driver Sep 14 16:43:19 Compiz isn't evil, if you set it up right. Things like "flaming windows" ARE stupid. Sep 14 16:43:32 what we're going for is a cairo backend worth a damn Sep 14 16:43:47 opengl accelerated gtk and firefox is just as good as wobbly windows Sep 14 16:44:19 How about "scale windows"? And "lamp" animation (if you even have a taskbar)? Sep 14 16:44:56 Granted, my netbook (not my primary system) barely has room for much other than one active window. Sep 14 16:45:23 Neko, yes, eth0 is fine, just not 127.0.0.1 or localhost Sep 14 16:45:26 And for now, unity is glitchy as all hell, both on intel and fglrx -- and kwin is glitchy on intel, as well. Spazzes and flickers white with every animation. Sep 14 16:45:29 there are simpler compositing systems that do apple expose-like fancy treats.. Sep 14 16:46:05 I also have menus open and close with "vacuum", and windows open with "dream" and close with "sidekick". Sep 14 16:46:14 That "sidekick" animation is surprisingly satisfying. Sep 14 16:46:41 we did have compiz working on a radeon 9200 with a 400MHz PPC once Sep 14 16:46:45 it worked ridiculously well Sep 14 16:47:08 wobbles and lamp minimize and the desktop cube and gravity and snap on windows.. Sep 14 16:47:26 * ogra doubts you will get compiz working on GLES Sep 14 16:47:36 what worries me is not the capabilities of the chip but the godawful compiz source Sep 14 16:47:39 That old "XGL" on top of GLES would be an interesting hack. Sep 14 16:47:42 but there is work going on to get unity/clutter working at least Sep 14 16:47:48 the cairo gles backend is disgusting Sep 14 16:47:56 Compiz 0.9 is slow, when 0.8 works fine. Sep 14 16:48:08 we have already fixed the evas backend though, so we could gl accelerate netbook-launcher-efl Sep 14 16:48:35 that would be sort of obtuse in a certain way but, it does work Sep 14 16:49:20 Some time later this week I'll try EGL on an R350. Sep 14 16:49:24 DanaG, compiz 0.9 they removed reliance on texture_from_pixmap in favor of a slower way that didn't depend on broken binary driver implementations.. Sep 14 16:49:52 0.9 gets 30 fps, where 0.8 gets 45-60. Sep 14 16:49:59 the idea being that you can copy a bitmap to a pci express 2.0 graphics card faster than you really need Sep 14 16:50:25 Last time I tried "mutter", the animations were confusing -- a minimize should NOT feel exactly the same as a close! Sep 14 16:50:35 what really screws it over though is X is very liberal about it's refreshing Sep 14 16:50:41 it just spams redraws that you don't need Sep 14 16:51:26 "Showrepaint" is useful, at least. Sep 14 16:52:04 do I mean liberal or conservative? I dunno. it sends more repaints than you functionally need, and it seems up to the driver to work it out... but if you wanted to optimize it you at least have to queue all these repaints and then work out what to do, which may cause some nasty lags Sep 14 16:52:46 QWS does the same thing, but then it's architecture is "subclass the paint engine and do it properly" and you have to do repaint management anyway, so.. Sep 14 16:52:50 X also does this stupid readback from video memory when allocating a new or resized window. Sep 14 16:53:02 I thought ajax fixed that Sep 14 16:53:07 like last week Sep 14 16:53:12 Ah, hadn't checked it lately. Sep 14 16:53:17 I hope that's true. Sep 14 16:53:35 oh wait no it may have been.. that guy with the name Sep 14 16:53:50 I found it was slow on radeon, intel, and nvidia (gf6150).... and slow as molasses on fglrx. Sep 14 16:54:02 my grey matter is saying Tejun Heo but that's the ata guy. Sep 14 16:57:38 I remember the discussion on it though and I am sure someone said they had fixed it Sep 14 16:58:39 I'm hung up (pun intended) on this tcflush fix ajax tried the other day though Sep 14 16:58:57 it stops my armel systems from using up 100% cpu in Xorg Sep 14 17:07:40 sigh Sep 14 17:07:47 ogra explain this approx thing again :/ Sep 14 17:08:34 it seems to just fail getting anything Sep 14 17:10:03 oh shit. typo. 9999 not 999 :) Sep 14 17:10:10 * Neko slinks back into a corner Sep 14 17:11:22 heh, so it solved itself ? Sep 14 17:12:04 MIRRORIP="http://$(ifconfig | grep inet | awk -F: '{ print $2 }' | awk '{ print $1 }' | grep --color=never 10.0.0):9999/ubuntu-ports" Sep 14 17:12:46 I just missed a 9. it's fine now. Sep 14 17:40:47 ooh that's annoyingly harmless Sep 14 17:40:57 makedev sets up /dev/agpgart on arm? Sep 14 18:58:35 eek Sep 14 18:58:36 debconf: apt-extracttemplates failed: Illegal seek Sep 14 18:58:36 Extracting templates from packages: 24%E: Could not open file /build/rootstock/tmp/linux-sound-base.template.54630 - open (2: No such file or directory) Sep 14 18:58:36 E: Unable to write to /build/rootstock/tmp/linux-sound-base.template.54630 - ofstream::ofstream (2: No such file or directory) Sep 14 18:58:36 E: Could not open file /build/rootstock/tmp/linux-sound-base.config.54631 - open (2: No such file or directory) Sep 14 18:58:39 ??? Sep 14 19:27:41 ogra why is rootstock packages trying to write directly into BUILDDIR for this stuff? Sep 14 19:45:14 argh it passes tmpdir into the VM? Sep 14 20:26:16 ogra, fixed the netbook-launcher-efl --add-favorite stuff Sep 14 20:26:23 patches attached in launchpad Sep 14 20:26:39 cause I am not sure of "proper procedure" around here, please be a little forgiving :) Sep 15 00:51:23 Finally managed to recover my nand on my beagleboard. Everytime I ran flash-kernel in Lucid, it would spew i/o errors writing the uInitrd. Erased it from uboot and reflashed from mmc. Works. Sep 15 00:51:48 Now testing Lucid->Maverick upgrade. Only 11 hours remaining. Whee. Sep 15 00:52:54 ouch :-) Sep 15 00:53:00 sd card is tooo slow :-( Sep 15 00:54:14 every task you have that include removing/updating/installing packages takes hours Sep 15 00:57:03 Luckily I have other platforms to do work on. But today has been platform hell day. **** ENDING LOGGING AT Wed Sep 15 02:59:57 2010