**** BEGIN LOGGING AT Wed May 07 02:59:56 2008 May 07 06:12:17 * * OE Bug 4195 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com May 07 06:12:19 * * diffutils_2.8.1.bb does not build on linux-uclibcgnueabi May 07 06:12:21 * * http://bugs.openembedded.net/show_bug.cgi?id=4195 May 07 06:12:29 * * OE Bug 4196 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com May 07 06:12:31 * * make 3.81 fails to compile on uclibc May 07 06:12:33 * * http://bugs.openembedded.net/show_bug.cgi?id=4196 May 07 07:32:32 coucou May 07 07:59:11 hi, I'm getting this with uptodate from .dev --> http://pastebin.com/m4730ecda May 07 08:31:47 cedric_: I see it too May 07 08:32:30 RP: ok May 07 08:34:52 03koen 07org.oe.dev * r3df1e17a... 10/ (1 packages/linux/linux-omap2_git.bb): linux-omap2 git: bump to HEAD May 07 08:34:56 03koen 07org.oe.dev * rd7b4b8c0... 10/ (4 files in 3 dirs): linux omap2 git: add L2 cache enable patch for beagleboard and update defconfig accordingly May 07 08:44:27 03rpurdie 07org.oe.dev * rea8b08f3... 10/ (1 packages/gnome/libsoup_2.2.100.bb): libsoup-2.2.100: Fix staging funciton to correctly handle the .pc files May 07 08:44:38 cedric_: try that ;-) May 07 08:57:53 RP: thx May 07 09:04:25 good moring everyone May 07 09:04:37 ehh morning I mean May 07 09:10:13 I wonder if anybody could help me figure out why my bitbake (first run) May 07 09:10:13 failed http://pastebin.com/m3f79efb7 May 07 09:11:35 exco: Fedora Core or RHEL May 07 09:11:54 morning May 07 09:12:01 or centos? May 07 09:12:03 morning hrw May 07 09:12:08 hrw: get that poky RedHat link ready :-) May 07 09:12:24 XorA: ;D May 07 09:13:07 XorA: Im running Ubuntu 8.04 if that's what you mean May 07 09:13:28 oh no, thats bad news May 07 09:13:49 What did they break? May 07 09:13:55 I've built the latest phoneme advanced! at last... May 07 09:14:37 cedric???? May 07 09:14:46 I did this laste year enough May 07 09:14:52 and then found midpath May 07 09:15:03 and decided phoneme advanced is crap May 07 09:15:36 no working qt port May 07 09:15:55 you cant change the vm May 07 09:16:20 echo: echo 0 > /proc/sys/vm/mmap_addr_min or something like that May 07 09:16:21 but thats my opinion May 07 09:16:32 4096 May 07 09:16:35 not 0 May 07 09:16:43 hrw: 0 works better May 07 09:16:48 RP: good to know May 07 09:17:04 hrw: 4096 leaves the change something else will fail, 0 means no limit May 07 09:17:13 s/change/chance/ May 07 09:17:22 ok May 07 09:17:45 ~curse slow svn repositories May 07 09:17:46 May the fleas of a thousand camels infest your most sensitive regions, slow svn repositories ! May 07 09:18:08 parsing of 787 recipes takes up to 8 minutes ;( May 07 09:18:24 uh May 07 09:18:32 mercurial *g* May 07 09:19:17 8 recipes to go.. May 07 09:20:10 hrw: There is a cache policy you can use to avoid that every time May 07 09:20:32 RP: not after change to configs iirc May 07 09:20:40 ok, build started May 07 09:21:15 exco: I've recently had the same problem on my Ubuntu machine May 07 09:21:30 but attributed it to me having OE stuff mounted over NFS May 07 09:21:43 hrw: BB_SRCREV_POLICY = "cache" May 07 09:21:53 as has paulproteus: http://pastebin.com/m12f68e84 May 07 09:21:57 RP: Is there something OE can do to fix this even when Ubuntu might be broken here? May 07 09:22:22 Laibsch: I'm planning on adding a check to poky so it warns if that value is != 0 May 07 09:23:04 RP: gracias May 07 09:23:08 Please add it to OE proper, too May 07 09:23:19 Laibsch: how do I attribute it ... so it works? May 07 09:23:29 attribute? May 07 09:23:41 "but attributed it to me having OE stuff mounted over NFS" May 07 09:23:47 ~hail git-svn May 07 09:23:48 * ibot bows down to git-svn and chants, "I'M NOT WORTHY!!" May 07 09:24:14 RP: Should we file a bug report against ubuntu, is this something that should be already set by ubuntu? May 07 09:24:30 exco: You misunderstand the meaning of attribute May 07 09:24:53 exco: rephrase "I believed it was failing because my OE stuff is on an NFS mount" May 07 09:25:07 RP: was echo 0 > /proc/sys/vm/mmap_addr_min addressed to me? May 07 09:25:41 Laibsch: Its a feature, not a bug. Its supposed to limit the potential for kernel security exploits May 07 09:25:44 It also breaks qemu :/ May 07 09:26:37 exco: I think so, yes May 07 09:26:46 Laibsch: yes, I got that wrong. But mine resides in folder on an ext3 partition May 07 09:26:48 RP: Then I guess we should fix up OE May 07 09:27:06 or upstream May 07 09:27:16 florian: Good morning May 07 09:27:31 Can you please unstick my mail sent to oe-devel? May 07 09:27:40 cat /proc/sys/vm/mmap_min_addr -> 65536 May 07 09:28:20 exco: Yes, it was directed at you. Setting that value to 0 will fix your problem May 07 09:28:40 Laibsch: Upstream will claim its not a bug May 07 09:29:46 RP: FWIW, that "file" does not even exist here May 07 09:29:53 Laibsch: ah sure, which one? May 07 09:29:57 Oh, but /proc/sys/vm/mmap_min_addr does May 07 09:30:11 florian: I sent you a mail yesterday night requesting the unfreeze May 07 09:30:20 thanks, RP ... so it bakes away (-: May 07 09:30:21 Nobody looking through the mod queue? May 07 09:30:33 Laibsch: right, just found your maikl in the inbox :) May 07 09:30:39 Good May 07 09:30:45 Laibsch: I was typing from memory hence "something like" ;-) May 07 09:30:47 hi florian May 07 09:30:50 Ah May 07 09:30:51 Sure May 07 09:30:59 Close enough ;-) May 07 09:31:04 morning, btw :) May 07 09:31:38 RP: If it is a feature as seen from ubuntu, not a bug that upstream will accept, can we fix it in OE or is a warning the best shot we have? May 07 09:32:38 Laibsch: You need to be root to change the value so we can't fix it that way. You can hack QEMU to work around the issue but that is suboptimal since QEMU isn't doing anything wrong as the fix impacts on performance May 07 09:32:57 I'm going to say we should therefore warn users to change the value May 07 09:32:58 you could add it to the wiki for Ubuntu installation May 07 09:33:52 * woglinde will give xine some love this evening May 07 09:34:26 OK, even if upstream will reject is as invalid, I'll shoot a bug their way nonetheless May 07 09:34:31 Qemu and? May 07 09:34:40 glibc? May 07 09:34:46 anything else? May 07 09:39:36 Laibsch: The only one who can help with the mail is mickeyl imho. May 07 09:39:36 Laibsch: It causes problems with qemu May 07 09:40:39 03rpurdie 07org.oe.dev * r057aae3f... 10/ (1 classes/sanity.bbclass): sanity.bbclass: Add check for problematic /proc/sys/vm/mmap_min_addr values May 07 09:40:52 RP: thanks May 07 09:41:16 RP: the glibc problem is a result of the qemu problem? May 07 09:41:23 and this is not only ubuntu May 07 09:41:23 Laibsch: yes May 07 09:41:36 Laibsch: can you resend the message, please? i was too slow to allow it and it's already autodeleted May 07 09:41:56 you guys autodelete the mod queue? May 07 09:42:01 Unbelievable May 07 09:42:46 say that again, when you are listed as owner for 10+ mailing lists and 99% of the incoming stuff is spam May 07 09:43:04 i get notified, so if it's a valid mail, i can react in one way or another May 07 09:43:26 mickeyl: It would be cleverer to reject these mails instead of just deleting the queue. May 07 09:43:45 mickeyl: The guy who runs the mailing lists for http://www.debconf.org/ has a pretty nifty automoderation system set up which takes care of almost all the spam in the moderation queues automatically. May 07 09:44:05 yes, there are a dozen of things one could do if one had time to do administration May 07 09:44:06 * florian points to his mail about hosting May 07 09:44:20 but alas, administrating stuff is not my primary business May 07 09:44:23 * XorA hosts his mail in florian May 07 09:44:31 mickeyl: If you were to ask him nicely he might even be willing to share it... May 07 09:44:38 * broonie goes and does that. May 07 09:45:09 mickeyl: This exactly is the point - we *must* split development from administration but everyone works into his own direction so we will never be able to do so. May 07 09:45:30 can OE generate enough funds for a sysadmin? May 07 09:46:05 uh May 07 09:46:11 XorA: Currently this is unlikely, but we need to search for peole who are interested in helping projects in this way. May 07 09:46:18 new job opurnity for me May 07 09:46:37 heh May 07 09:46:45 it's my hope that with the e.V. in place, we can raise enough funds to hire an admin May 07 09:46:57 I think it should be clear that I'd be willing to help out admin'ing May 07 09:47:04 mickeyl: that would rock May 07 09:47:12 yeah May 07 09:47:15 But only on close to 100% packaged system May 07 09:47:20 preferably ubuntu May 07 09:47:24 cause things just work May 07 09:47:28 * woglinde updates the c.v. May 07 09:47:33 and are reasonably recent May 07 09:47:49 * broonie has had some "fun" with Ubuntu security updates for universe :( May 07 09:47:50 did not nslu2 guys offered admin help? May 07 09:47:52 mickeyl: I'd take the money if there was some on the table ;-) May 07 09:48:02 * XorA downgraded ubuntu in his preferences May 07 09:48:21 broonie, XorA: please elaborate May 07 09:48:37 Laibsch: ubuntu seems to be going the way of redhat, crazy patches for no real gain May 07 09:48:46 example May 07 09:48:48 ? May 07 09:48:49 Laibsch: lenny fits my needs better than Hardy Heron May 07 09:49:00 Laibsch: cups, its pretty much unusable on ubuntu May 07 09:49:04 Laibsch: A few broken packages - the worst was a Mailman update which wouldn't actually run, breaking postinst unrecoverably. May 07 09:49:22 well, look at amethyst wrt lenny, nothing but hand-installed stuff already because everything is too old May 07 09:49:30 Laibsch: compulsary compiz, which fucks my gfx up May 07 09:49:38 broonie: when was that? breezy? May 07 09:49:41 They did put out a fixed version fairly sharpish but it's *not* what I'm looking for from security updates. May 07 09:49:53 /my upgrades have ever since been so smooth May 07 09:50:07 Laibsch: lenny is too old? its not even released May 07 09:50:15 Laibsch: Nope, gutsy. Like I say, they did fix it promptly. May 07 09:50:22 mickeyl: oh right, we need to make sure we get the e.V. ready May 07 09:50:35 With debian (few years ago), I had breakage (severe breakage) whenever I did "aptitude update". Yes, every f***ing time May 07 09:50:41 i sincerely hope zecke will do the necessary preparations to make it happen on LT May 07 09:50:47 or rather, during LT May 07 09:50:54 Laibsch: On stable? May 07 09:51:03 mickeyl hm May 07 09:51:10 Can you really run a desktop system on stable? May 07 09:51:19 With no packaging and backporting of your own? May 07 09:51:23 * Laibsch does not think so May 07 09:51:24 * XorA has actually done an aptitude dist-upgrade to switch debian to ubuntu, then went back again, that was freaky but cool May 07 09:51:33 scary May 07 09:51:35 !!! May 07 09:51:36 woglinde: can you kick him? :) May 07 09:51:57 * broonie has done that (not at present, but then as a debian developer there's this whole development thing going on...). May 07 09:52:02 mickeyl I kick him all the time that he does his last oral exam May 07 09:52:06 for vordiplom May 07 09:52:11 broonie: Well, I guess every distro makes mistakes and universe is not official May 07 09:52:12 yes, that's important as well May 07 09:52:16 The only backporting thing I've really needed is git. May 07 09:52:19 Which I think is a good concept May 07 09:52:33 Laibsch: I'm not sure ametyst is lenny, is it not etch? May 07 09:52:44 Laibsch: Indeed, but it does meant hat you're running more of a risk once you need stuff outside main. May 07 09:54:29 florian I think I can man the LT booth the whole friday May 07 09:54:35 XorA: cups works just perfect here. sorry about lenny, I meant whatever stable version is currently installed on amethyst (I am not so current with the debian naming). Whenever a new service needs to be installed on amethyst you discover that either no package or an outdated one is available May 07 09:54:49 mickeyl: Looking at the wiki and the reactions to my latest LinuxTag mail I would strongly recommend a day before or after. May 07 09:55:10 florian I just update the wiki now May 07 09:55:10 woglinde: cool, that's good news... the first of this for a logn time now :) May 07 09:55:24 Laibsch: it works for you, that is not the same as it works for everyone, try some networking with it, and see how fucked ubuntu made it May 07 09:55:32 florian but I have to ask my wife it will be okay May 07 09:55:41 In any case, I'm just sharing my preference as one who did a lot of admin'ing lately May 07 09:55:42 and their samba conf is fucked as well May 07 09:55:59 XorA: cups? I do networking May 07 09:56:17 XorA: dito, samba works just fine May 07 09:56:21 The way I want it May 07 09:56:35 not the way windows machine want it though May 07 09:56:41 of course May 07 09:56:56 My GF has windows, even a Japanese version which makes it even harder May 07 09:57:10 hm May 07 09:57:25 * broonie manages to get Ubuntu samba to interact with the Windows machines at work (client only, though). May 07 09:57:38 I use debian now over 10 years and it doesnt give me the pain a suse or redhad did May 07 09:57:47 Laibsch: my friend once had to install japanese windows without knowledge of japanese.. May 07 09:57:48 * Laibsch agrees ;-) May 07 09:58:10 I use Debian and refuse to use Ubuntu or other derivative May 07 09:58:12 hrw: Just installing you can use your "graphical experience" May 07 09:58:24 hrw: Try troubleshooting when something does nto work ;-) May 07 09:58:29 Then it becomes another story May 07 09:58:58 florian I hope I can bring my simpad May 07 09:59:06 to the linuxtag May 07 09:59:07 Again and in any case, I'm just sharing my preference as one who did a lot of admin'ing lately for OE infrastructure May 07 10:00:14 my "graphical experience" with Windows past Win98 (which is when I stopped using it) has also declined considerably May 07 10:01:17 Laibsch: FWIW if you're having trouble with out of date stuff on debian backports.org and/or volatile may be of help too (both are moving into being part of the official Debian infrastructure). May 07 10:01:26 I know May 07 10:01:32 hrw: installing Japanese windows is no harder than normal windows, Ive done it many times May 07 10:01:42 woglinde: I've tried jamvm, cacao on my device (pxa250, 32M) and it's fine for command line program but to just display a simple UI --> not enough memory/too slow May 07 10:01:56 broonie: But they also don't have the stuff we need (more often than not, at least). backports.org is already in sources.list, IIRC May 07 10:02:09 mickeyl: Can you remember which packages caused you to make autotools.bbclass staging .pc files opt in? May 07 10:02:15 cedric did you try cacao-clcd too? May 07 10:02:26 Laibsch: Interesting. Which version of Debian are you using and what sort of stuff are you using? May 07 10:02:27 woglinde: A working Simpad would be good... my Simpad is in some undefined state :) May 07 10:02:28 woglinde: nope May 07 10:02:31 mickeyl: I'm thinking it should really be opt-out :/ May 07 10:02:37 broonie: amethyst? May 07 10:02:51 Laibsch: Where you're having trouble, yes. May 07 10:02:53 cedric what is your goal? May 07 10:03:04 Laibsch: To get amethyst to use packages from backports, you have to tell it explicitly May 07 10:03:09 cedric I dont believe that opera-mini will run with phoneme May 07 10:04:04 broonie: http://rafb.net/p/uI9Z2a58.html is for amethyst May 07 10:04:19 woglinde: run SVG (JSR 226) and doing some UI prototyping (AWT, SWT, whatever) May 07 10:04:30 cedric hehe May 07 10:04:35 then look at midpath May 07 10:04:41 RP: aptitude will resolve with backports packages if they are available May 07 10:04:58 and cacao-clcd May 07 10:05:05 RP: Do we use pinning? or why is backports.org not installed by default IYO? May 07 10:05:07 Laibsch: you do need to configure things to pull in backports packages if there is a version in stable. May 07 10:05:16 03koen 07org.oe.dev * r516e70a5... 10/ (5 files in 2 dirs): May 07 10:05:16 libsoup: modify staging funciton to not symlink, we don't want libsoup.pc, but libsoup-2.x.pc May 07 10:05:16 epiphany: depend on libsoup-2.4 May 07 10:05:38 I am fighting with some qtopia bugs, but for qt4,gtk,awt and fb it runs fine May 07 10:05:39 Laibsch: That'll be why - the default pinning will prefer stable over backports. May 07 10:05:43 broonie: My understanding of apt and pinning is different May 07 10:05:54 Check again :) May 07 10:06:07 We don't have pinning May 07 10:06:14 woglinde: which device? for SVG I need to wait for midpath 0.3 May 07 10:06:31 cedric I am playing with the greenphone May 07 10:06:49 cedric you can check out the svn May 07 10:06:51 it works May 07 10:08:17 woglinde: so with cacao-clcd I can use the gtk peer aswell? can we go offline (out of the OE channel) to discuss a little bit more pls? May 07 10:09:32 RP: the major reason for it to be opt-in was that carefulness May 07 10:09:40 RP: i agree that it should be opt-out May 07 10:09:53 Laibsch: There's a default May 07 10:13:44 OK, can we end this discussion? I think debian is a nice system. I prefer ubuntu. I'd help admin'ing a Debian system if we can ensure that the software on it remains mostly packaged. My experience with ltg and even amethyst already is that debian does not allow that because nobody wants to go to testing or unstable. May 07 10:14:32 I believe the experience with ltg and amethyst show that debian stable cannot do it for an OE server. It might very well be that ubuntu can't either. I have better hopes, though. May 07 10:15:12 Sorry for inviting another round of discussions ,-) May 07 10:15:45 I'm as interested in finding out what you're missing with my Debian hat on since it's useful stuff to feed into "discussions" about what we're doing with stable releases and maintaining them. May 07 10:16:00 broonie: bugzilla 3.x for example May 07 10:16:13 hrw: Thanks, I was just about to raise that May 07 10:16:19 mickeyl: How do you think we should switch to that? May 07 10:16:21 But to be fair, I am not sure it is in ubuntu May 07 10:16:41 broonie: RP mentioned some git stuff May 07 10:16:44 mickeyl: I'm thinking we should perhaps phase out pkgconfig.bbclass entirely... May 07 10:16:55 Laibsch: it is not - as usual May 07 10:17:03 RP: ping? May 07 10:17:04 broonie: koen has already hand-rolled mtn again (which is already leading to problems, it seems) May 07 10:17:25 broonie: Just take any service, really May 07 10:17:26 AFAICT our bugzilla packages are a bit of a disaster zone :/ May 07 10:17:28 woglinde: have you got a svn midpath bb file in your local repository? May 07 10:17:33 Except for apache and friends, maybe ;-) May 07 10:17:37 Laibsch: show me packages which are in newer versions in ubuntu (and those versions do not exists in experimental) then in debian May 07 10:17:45 RP: i agree. for the few packages that do not automatically install the pkgconfig files we can hand-code that in staging May 07 10:18:03 hrw: Are you willing to run experimental on the server. That is exactly my poiint. May 07 10:18:11 git is definitely up to date in backports (I use it from there); mtn I'd be shocked if it were given that OE is close to the only user :) May 07 10:18:21 broonie: git stuff, not git itself May 07 10:18:24 Laibsch: there is no such thing as experimental, you can cherry pick from it, but not run it May 07 10:18:27 Ask RP what it was May 07 10:18:28 Laibsch: having unstable version of glib by default (ubuntu way) is not sane too May 07 10:18:40 It is supported May 07 10:18:54 which is what I care about in the server distro May 07 10:19:07 RP: Laibsch mentioned that there was some git stuff you couldn't find packaged for etch but couldn't remember what it was? May 07 10:19:10 I want to offload the admin'ing responsibility to the distro maintainer May 07 10:19:26 Laibsch: Thanks, that's all useful information. May 07 10:19:36 My feeling is that it works better with ubuntu or at least is a chance. FOr several reasons it did not work with debian May 07 10:21:14 XorA: the point remains, right? Do you remember what got ltg into serious trouble? So much that a simple "aptitude update" bombed out until *I* fixed it. A problem that had lead to several rounds of security updates not installed? May 07 10:21:44 s/lead/led/ May 07 10:21:45 cedric_ can you send private messages? May 07 10:21:51 Laibsch: that smells of people not knowing how to use apt than a fault with a distro May 07 10:21:59 both May 07 10:22:11 * XorA has used hold and dpkg ability to fake packages to keep servers sane before May 07 10:23:32 the problem is that there are few persons which touch apt(itude) May 07 10:23:52 there should be one with apt power and rest should give him requests May 07 10:23:53 * Laibsch hugs aptitude May 07 10:24:01 aptitude to the rescue ;-) May 07 10:25:12 broonie: gitosis, cgit May 07 10:25:29 apitude is "new" crap May 07 10:25:34 broonie: gitosis is more available now I think, I built my own packages for that... May 07 10:25:36 dselect is the reel tool May 07 10:25:44 yuck May 07 10:26:05 RP: Hrm. cgit doesn't even seem to be packaged.... May 07 10:26:52 RP: I have been playing around with x11 on qemux86 and was wondering: would there be any problem build mousedev and psmouse into the kernel? May 07 10:26:58 RP: gitosis is packaged at least so backports.org should be trivial. I'll have a look and maybey prod daniel. May 07 10:27:34 RP: currently they are provided as modules, which means depending on + loading the kernel modules, how do you handle it up until now? May 07 10:27:39 cedric_ sorry I have to go now for lunch, maybee we can talk later or theis evening May 07 10:27:48 bye May 07 10:28:31 florian: Did you make the planet.ltg.org change? May 07 10:28:31 jeremy_laine: I use it with the wacom touschscreen emulation May 07 10:28:52 jeremy_laine: See poky's qemu scripts... May 07 10:28:54 Laibsch: oops forgot to mention - yes I did. May 07 10:29:03 florian: Thanks May 07 10:29:05 jeremy_laine: That way it doesn't have to grab the cursor May 07 10:29:16 RP: what do I need to install to get this to work? do you use xserver-kdrive-vesa? May 07 10:29:18 florian: Please correct the URLs, though ;-) May 07 10:29:18 Laibsch: yw May 07 10:29:25 woglinde: ok so after lunch, this evening will be not possible for me... May 07 10:29:45 jeremy_laine: Probably tslib. Are you using kdrive or not? May 07 10:30:02 broonie: gitosis in backports would rock :) May 07 10:30:16 cedric_ okay can promise because we have administration day at the university and the computers in my department may out of order May 07 10:30:17 broonie: There is another server I could use that on too ;-) May 07 10:30:18 RP: yep, so far kdrive/vesa and using "-usbdevice mouse" for qemu May 07 10:30:35 (for some reason the PS/2 mouse in unresponsive) May 07 10:31:07 jeremy_laine: http://svn.o-hand.com/view/poky/trunk/scripts/poky-qemu-internal?rev=4437&view=markup May 07 10:31:09 woglinde: :-) May 07 10:31:30 jeremy_laine: -usb -usbdevice wacom-tablet May 07 10:31:44 re May 07 10:31:48 RP: I will check what you asked for now May 07 10:31:51 jeremy_laine: Try one of poky's images, it all just works there May 07 10:32:02 RP: is the wacom driver built into the kernel? May 07 10:32:18 jeremy_laine: in poky yes, so probably in OE too May 07 10:32:37 RP: also, the console seems to get messed up (by psplash I guess) so I need to put my console on ttyS0! May 07 10:33:18 jeremy_laine: Which kernel version and framebuffer are you using? May 07 10:33:33 The vesa stuff in recent qemus/kernels caused me a few problems... May 07 10:33:52 RP: 2.6.24, video=vesafb May 07 10:34:30 jeremy_laine: linux-rp? May 07 10:34:43 RP: yup, the one and only :) May 07 10:35:21 I have to admit I've only really used 2.6.23 May 07 10:36:18 RP: heh, that sucks 2.6.24 is default for angstrom :) May 07 10:36:41 jeremy_laine: I think the simplest thing to do might be to try a build with poky since its known to work, get that working, then compare the bits with OE :/ May 07 10:37:05 I'll merge anything needed to OE but I'm totally lost off with where things are at sync wise May 07 10:37:31 RP: will try 2.6.23 first for now, thanks for the pointers! May 07 10:38:10 RP: if it turns out 2.6.23 fixes things I might edit linux-rp_2.6.24.bb and remove the preferred version for qemux86 May 07 10:39:34 jeremy_laine: ok, let me know if that is the issue, just so I'm aware of it! :) May 07 10:40:02 RP: here are the answer to your questions: http://pastebin.mozilla.org/424841 and I am not mixing uclibc/glibc builds, started with a clean build (i.e. wiped temp) May 07 10:42:41 Jin^eLD: arm-angstrom-linux-uclibcgnueabi/usr/sbin/ldconfig sounds suspicious May 07 10:43:28 Jin^eLD: Can you run file against that please? May 07 10:44:08 arm-angstrom-linux-uclibcgnueabi/usr/sbin/ldconfig: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), not stripped May 07 10:44:35 RP: grr, 2.6.23 and 2.6.24 give me the same thing: I get the touchscreen calibration screen but when I click nothing happens May 07 10:44:39 ah, its from uclibc which has the string "libc.so.6" embedded in it? :/ May 07 10:45:05 jeremy_laine: Try your kernel with a poky image? May 07 10:45:33 let me check May 07 10:46:06 yeah, it has the libc.so.6 May 07 10:47:00 RP: I guess but that means setting up another build env :) May 07 10:47:18 jeremy_laine: You can pull prebuilt images May 07 10:47:41 RP: will revert to USB mouse for now I think, would there be any adverse effect to building mousedev and psmouse into the kernel? May 07 10:47:45 http://pokylinux.org/autobuild/poky/ May 07 10:48:21 jeremy_laine: As I said before, using a mouse will mean you are forced into screen grabs May 07 10:49:30 RP: what I don't understand is how the touchscreen changes this, whenever I click the qemu window I get grabbed? May 07 10:49:51 jeremy_laine: In touchscreen mode it doesn't do that May 07 10:50:20 jeremy_laine: mice only transmit relative motion so there is no way to tell qemu an absolute position May 07 10:50:33 jeremy_laine: with a touchscreen you deal with absolute position so its much easier May 07 10:51:46 re May 07 10:51:56 RP: well, seems like glibc has not been staged or built or whatever May 07 10:52:15 Jin^eLD: ok. Has uclibc staged? May 07 10:52:49 ./sulu-angstrom-linux-uclibcgnueabi/uclibc-0.9.29+svnr21926-r7.do_populate_staging May 07 10:52:59 ./sulu-angstrom-linux-uclibcgnueabi/uclibc-0.9.29+svnr21926-r7.do_package_stage_all May 07 10:53:06 are those the tasks? May 07 10:53:18 did it stage before or after the gcc-cross-sdk started ? May 07 10:53:22 yes, its staged May 07 10:53:59 probably before... I think the -sdk packages only build after the gcc-cross and binutils-cross are build, and gcc-cross requires uclibc May 07 10:54:48 If you rebuild gcc-cross-sdk does it work second try? May 07 10:55:03 I'm trying to think of the race involved here May 07 10:55:10 if it is a race... May 07 10:55:30 Approach two is try running the command thats failing in gcc-cross-sdk manually May 07 10:55:31 RP: -c rebuild binutils-cross-sdk consistently produces the libc.so.6 shared library provider message May 07 10:55:59 ok, can you open up a devshell and manually run the command and produce the failure? May 07 10:55:59 and gcc-cross-sdk is a different story, it either fails completely like in the log, or it builds but also produces the message May 07 10:56:20 I did not do much stuff with the devshell yet, bitbake -c devshell gcc-cross-sdk ? May 07 10:56:34 ah, so binutils-cross is problematic? May 07 10:56:39 binutils-cross-sdk is problematic May 07 10:57:01 broonie: Are you a DD? May 07 10:57:02 ah, but that message is normal May 07 10:57:19 Laibsch: Yes. May 07 10:57:19 RP: my suspicion is that gcc-cross-sdk fails to build because of some badness in bintils-cross-sdk May 07 10:57:25 poky will do that as well for binutils-cross-sdk May 07 10:57:46 Jin^eLD: No, that message is normal for binutils-cross-sdk May 07 10:57:47 broonie: Are you interested in sponsoring my packages by any chance? http://mentors.debian.net/cgi-bin/sponsor-pkglist?action=details;package=gourmet and http://mentors.debian.net/cgi-bin/sponsor-pkglist?action=details;package=pastebinit May 07 10:57:50 RP: with binutils-cross and gcc-cross, in both, poky, oe stable and oe dev - you do not have the "mssing shared lib provider bleh" message May 07 10:58:00 in poky you also do not have this message for the -sdk packages May 07 10:58:03 broonie: Hehe, gotcha ;-) May 07 10:58:06 but in .dev and .stable you do have this message May 07 10:58:29 Jin^eLD: What is means is that its noticed the binutils binaries link against libc.so.6 but it can't find any package which provides this May 07 10:58:51 Laibsch: Meh. Could you drop me an e-mail at broonie@sirena.org.uk and I'll try to at least have a look? May 07 10:58:53 RP: well isn't that part of the problem? why are they linking against libc.so.6 - I am doing a uclibc build May 07 10:58:55 Jin^eLD: That is entirely expected since they're binaries meant to run with the system libc but we don't package the system libc since that makes no sense May 07 10:58:59 Do you follow? May 07 10:59:06 broonie: Sure May 07 10:59:08 aha, so that's for the native libc? May 07 10:59:12 but why is this message not there on poky? May 07 10:59:21 Jin^eLD: You want to run the binutils binaries on the build system, not the target May 07 10:59:46 You don't see this for bintils-cross since it never packages May 07 11:00:01 RP: no, in poky I also do not see that message for binutils-cross-sdk May 07 11:00:10 that's why I was suspicious May 07 11:00:31 if you say that this is not related to the problem and that its only for the system libc - then I was on the wrong trail May 07 11:01:30 Jin^eLD: I just ran "bitbake bintils-cross-sdk -c package -f" and I see those warnings here May 07 11:01:34 Jin^eLD: It does happen in Poky and its harmless May 07 11:01:44 its not related to your other issue May 07 11:02:04 hmm OK.. I wonder why I did not see that messages, but well, if its not realted to the problem then whatever :) May 07 11:02:21 Jin^eLD: You'll just have missed them, they'll be there May 07 11:02:29 could be May 07 11:02:31 :) May 07 11:03:18 Jin^eLD: Try and get the error reproduced under a devshell May 07 11:03:22 ok May 07 11:03:40 Jin^eLD: If you can do that, you can use strace to try and find out where the libc.so.6 is coming from May 07 11:03:55 ok, good idea May 07 11:04:25 do you have a link or a hint on the devshell usage? I got the devshell terminal now May 07 11:04:27 what do I do with it? May 07 11:04:54 You have the failed compile log? May 07 11:05:09 cd into the directory where it failed, then run the command that failed May 07 11:05:31 ok, so I first let it fail, then do -c devshell and then repeat the last command, right? May 07 11:05:43 yes May 07 11:05:47 ok let me do that... May 07 11:06:06 the idea being to examine the failure more closely May 07 11:06:33 RP: the poky kernel + image seems to give me a functional X, except I don't see my cursor May 07 11:06:51 RP: how do you know where you're about to click? May 07 11:07:22 jeremy_laine: Which qemu binaries are you using? May 07 11:08:09 tried with Debian/unstable's qemu and kvm May 07 11:08:27 jeremy_laine: There are tweaks in Poky for that, or use the "-show-cursor" option May 07 11:08:48 jeremy_laine: I'd recommend the Poky built qemu binaries in a way May 07 11:08:59 although its more important for ARM than x86 I guess May 07 11:11:51 RP: very cool, thanks a lot! works like a charm on kvm (am using the SDK image) May 07 11:12:38 just need to get my keyboard mapping right now.. (azerty) May 07 11:22:37 jeremy_laine: :) May 07 11:32:36 can i use jffs2 rootfs image with qemu? May 07 11:32:43 RP: ok, I reproduced the build problem, now will try with devshell May 07 11:41:10 I have something interesting for all the "packaged software"-lovers out there May 07 11:41:10 /me presents apt-get'able Openembedded: http://blog.leggewie.org/?p=39 May 07 11:41:44 RP: I got the strace of the failure, trying to find something in it now May 07 11:42:00 hrw: Do you think that is something for the PR department? May 07 11:42:40 Laibsch: what exactly is being apt-getted? May 07 11:42:44 the metadata? May 07 11:42:54 Jin^eLD: If that doesn't help, it may pay to increase the string sizes, there is some option to strace iirc May 07 11:43:20 -s 0 May 07 11:43:21 RP: right now I am not even sure what I am looking for in the trace May 07 11:43:35 Jin^eLD: I'll eventually commit the definition to OE itself May 07 11:43:38 RP: but - if I remove the -lc at the end of the compile line May 07 11:43:45 then it does link May 07 11:43:54 Jin^eLD: The sources are included for download if you want to take a look May 07 11:43:59 with -lc I have the libc.so.6 error May 07 11:44:08 Jin^eLD: It is an extension of openembedded-essential May 07 11:44:12 Jin^eLD: Can you share the strace output? (compress it) May 07 11:44:20 RP: one sec May 07 11:45:47 http://www.deadlock.dhs.org/jin/sdktrace.txt.bz2 May 07 11:45:55 or 29k uncompressed http://www.deadlock.dhs.org/jin/sdktrace.txt May 07 11:46:04 its not much, I traced only the last failing command May 07 11:46:30 Jin^eLD: OE-essential: pull in all packages required to build May 07 11:46:30 Jin^eLD: OE-common: pull in all packages required to build, pull and update meta-data May 07 11:46:30 Jin^eLD: OE-distro: depend on OE-common, pull in more stuff for distro if needed, include shell script to set up environment for OE compilation with sane defaults, configurable via environment var and /etc May 07 11:46:46 I think that is it May 07 11:46:53 Laibsch: sounds cool May 07 11:46:54 I may have forgotten something May 07 11:47:09 But it is really "install and use immediately" May 07 11:47:46 OE resides in /usr/src/oe and the user compiles and can also add personal stuff like additional recipes in ~/oe May 07 11:47:56 RP: http://www.deadlock.dhs.org/jin/failing_command.txt this is the command for which I did the trace, note the -lc in the end, if I remove it it will not cause an error, with -lc I get the libc.so.6 error, not sure how relevant that is though May 07 11:48:57 hmm, maybe I should use -ff, I think it forks off some stuff May 07 11:49:24 yeah, the strace is not complete, there are 3 more processes May 07 11:49:28 Jin^eLD: use strace -f to catch the trace for the children May 07 11:49:40 Jin^eLD: of -ff May 07 11:49:51 but you're ahead of me :) May 07 11:49:55 :) May 07 11:51:14 RP: http://www.deadlock.dhs.org/jin/trace/ May 07 11:51:53 rwhitby: Are you interested in creating openembedded-openmoko? Should be very easy, but I have no experience with the steps to set up an OE environment. It would render the problems with the Makefile moot. The package I created is basically a scripted and packaged GettingStarted. bitbake and friends are never hidden. May 07 11:52:11 s/OE environment/Openmoko environment/ May 07 11:52:45 Laibsch: what exactly is openembedded-openmoko? May 07 11:53:11 Have you read http://blog.leggewie.org/?p=39 May 07 11:53:13 ? May 07 11:53:20 Jin^eLD: Notice it reading from /opt/angstrom/arm/arm-angstrom-linux-uclibcgnueabi/lib/libc.so May 07 11:53:22 My requirement for a build system is that I should be able to start a shell, cd into a directory, and type a single command to build the final image from scratch. May 07 11:53:29 Jin^eLD: What toolchain do you have there? May 07 11:53:43 rwhitby: I believe that should be possible May 07 11:54:00 except, you don't cd intothe directory May 07 11:54:08 you call oe-setup-openmoko May 07 11:54:13 rwhitby, Laibsch: poky ;-) May 07 11:54:13 RP: wow... I wonder why it is doing so? is the "extrnal toolchain" kicking in? May 07 11:54:14 Then "bitbake $image" May 07 11:54:25 Laibsch: that's two commands :-) May 07 11:54:32 RP: openembedded-poky! ;-) May 07 11:54:47 the cd is important, cause I have a directory for slugos, a directory for openmoko, etc ... May 07 11:54:54 Jin^eLD: I think having a broken toolchain on the path where the new one will end up is breaking things May 07 11:55:11 rwhitby: If you insist, you could have another scripts to call those two commands in sequence May 07 11:55:23 RP: the difference seems to be in the rootfs, not the kernel (for the qemu / x11 issue) May 07 11:55:40 rwhitby: The system is set up so that different OE-derived distros don't step on each other's toes May 07 11:55:41 RP: it is not in path though... May 07 11:55:46 RP: kernel linux-rp_2.6.23 works with the poky image May 07 11:55:58 rwhitby: OE resides in /usr/src/oe and the user compiles and can also add personal stuff like additional recipes in ~/oe May 07 11:56:06 RP: if I look at the path in devshell - this toolchain I got in /opt is not in $PATH May 07 11:56:12 Laibsch: also, what if I'm building on CentOS, not Debian? May 07 11:56:15 Jin^eLD: I don't mean PATH, I mean in the place the toolchain is destined for May 07 11:56:25 rwhitby: CentOS has debs? May 07 11:56:31 no May 07 11:56:33 If not, you are out of luck May 07 11:56:34 Laibsch: alient works both ways! May 07 11:56:36 RP: I see... that's a caveat I guess, let me remove it and see if it helps May 07 11:56:46 broonie: I am fine with people trying that May 07 11:56:58 Laibsch: that seems to go against the OE idea of host platform independence then ... May 07 11:57:00 broonie: I'll help out but I won't guarantee it'll work May 07 11:57:11 RP: you are right! that was it! May 07 11:57:17 rwhitby: Nobody said anything about requiring people to use it May 07 11:57:24 after rm -rf /opt/angstrom the failing compile line actually compiled May 07 11:57:25 It just makes it a lot easier May 07 11:57:44 RP: we should somehow document that :) May 07 11:57:49 rwhitby: If you don't see the beauty of this, I don't want to force you. May 07 11:57:57 RP: what provides vm86d in poky? opkg search vm86d doesn't give me anything May 07 11:57:57 I thought it was a no-brainer May 07 11:58:15 Laibsch: agreed. It just doesn't meet my requirements. You were asking whether I was interested in it. I answered from my personal viewpoint. I don't try and make others share my personal viewpoint :-) May 07 11:58:37 RP: and btw, the libc.so.6 fix in meta-toolchain (removing the GROUP thing) is only in .dev, so its still broken in .stable; of course I can easily work around this now, but someone else may still stumble upon this May 07 11:58:37 It is indeed a great idea. May 07 11:59:16 Jin^eLD: Thats something for the stable branch maintainers to deal with May 07 11:59:24 Laibsch: I'm not discounting it - just discussing the pros and cons. May 07 11:59:35 We should probably warn if the target toolchain path exists... May 07 11:59:46 rwhitby: I think I'll try to create openembedded-openmoko nonetheless. Looking forward to get your (coerced) support despite your hesitation ;-) May 07 12:00:17 RP: warn or even exit May 07 12:00:20 jeremy_laine: There is a package there, meta-packages/v86d/v86d_ May 07 12:00:29 Jin^eLD: yes, error might be best May 07 12:00:29 RP: since the results may be random depending on what stuff you have in there May 07 12:00:38 Laibsch: I can see the benefit of it, and may well decide to help :-) May 07 12:00:39 that cost me about a week, doh :> May 07 12:00:53 rwhitby: I just need to know for example if openmoko uses straight org.oe.dev or not. After reading the wiki yesterday, I was still unsure about it. May 07 12:01:17 Laibsch: it's all encoded in the MokoMakefile - just read that. May 07 12:01:26 (it's an executable specification of the build environment) May 07 12:01:30 * Laibsch does not understand Makefiles very well May 07 12:01:50 I had a hard time putting together the debian/* files which are also makefiles May 07 12:02:00 debian/* files are weird May 07 12:02:04 learning by example May 07 12:02:05 yes May 07 12:02:20 My makefiles should be easier to understand :-) May 07 12:02:28 but dh_make helped, as did #debian-mentors May 07 12:02:42 rwhitby: I'll just ask you in plain english May 07 12:02:44 ;-) May 07 12:03:03 Laibsch: and I'll point you to a line number in the Makefile ;-) May 07 12:03:21 So, again, is it org.oe.dev from monotone.oe.org or do I need to check out a 99% identical copy from monotone.om.org? May 07 12:04:00 Well, I don't have any benefit from openmoko, I don't own a device May 07 12:04:10 So, I might as well drop it any time May 07 12:04:40 Laibsch: your out of date, openmoko is a district branch now May 07 12:04:57 XorA: I never was uptodate ;-) May 07 12:05:03 Laibsch: it's org.openmoko.dev from monotone.openmoko.org May 07 12:05:11 * XorA aptitude udate Laibsch May 07 12:05:23 (see setup-openembedded rule in the Makefile) May 07 12:05:27 broken dependency May 07 12:05:42 next version requires CS knowledge ;-) May 07 12:11:52 RP: will you take care of error'ing out or shall I submit something additional? May 07 12:12:01 i.e. PR or change request or whatever? May 07 12:12:07 RP: sweet, kernel-module-uvesafb + v86d works in OE May 07 12:12:19 RP: I'll import v86d into OE, OK? May 07 12:14:24 jeremy_laine: yes May 07 12:19:38 03jeremy_laine 07org.oe.dev * r6f6193f9... 10/ (7 files in 4 dirs): v86d: import v86d (from poky) for uvesafb support May 07 12:20:04 well, I added the info we found out to the PR and lowered the Importance since a workaround now exists. I hope it gets fixed in .stable, all information and solution is in the PR May 07 12:20:11 RP thanks again for helping out! May 07 12:35:12 03jeremy_laine 07org.oe.dev * r5eef2a39... 10/ (4 files in 4 dirs): pointercal_0.0.bb: add pointer calibration for qemux86 (from Poky) May 07 12:40:41 RP: what would you think of adding "touchscreen" to the MACHINE_FEATURES in conf/include/qemu.inc ? May 07 12:46:27 jeremy_laine: Given the way we use them thats probably appropriate May 07 12:47:06 RP: with this change I have gpe running on OE / qemux86 nicely May 07 12:47:25 RP: with the latest linux-rp 2.6.24 kernel May 07 12:48:46 RP: maybe even add MACHINE_EXTRA_RRECOMMENDS = "kernel-module-uvesafb v86d" for qemux86 May 07 12:51:38 nicely done: May 07 12:51:38 /proc/sys/vm/mmap_min_addr is not 0. This will cause problems with qemu so please fix the value (as root) May 07 12:51:42 thanks May 07 13:01:35 what is the difference between openembedded, openmoko, and poky metadata ? Asked differently, what is in openembedded but not in openmoko metadata, and what is in poky but not in opembedded ? May 07 13:03:00 * cyrilRomain notices openmoko branches includes openembedded branches, but maybe outdated one since there is org.openembedded.angstrom-2007.12-stable and not org.openembedded.stable May 07 13:04:19 poky is a stable oe branch/forks for some chosen packages (with syncs between themselves), isnt it? May 07 13:04:56 afaik Poky is a subset of OE packages, but stable and only the stuff that is needed there May 07 13:05:51 03jeremy_laine 07org.oe.dev * rcdaae06c... 10/ (1 conf/machine/include/qemu.inc): qemu.inc: add touchscreen to MACHINE_FEATURES May 07 13:05:55 03jeremy_laine 07org.oe.dev * r152bd050... 10/ (1 conf/machine/qemux86.conf): qemux86.conf: add MACHINE_EXTRA_RRECOMMENDS for framebuffer support May 07 13:07:31 jeremy_laine: Do you think that might fix May 07 13:07:36 !oebug 3742 May 07 13:07:37 * * Bug 3742, Status: NEW, Created: 2008-02-05 14:08 May 07 13:07:38 * * : Latest opie images fail calibration (in qemu) May 07 13:07:39 * * http://bugs.openembedded.net/show_bug.cgi?id=3742 May 07 13:08:07 ~curse OE wiki May 07 13:08:09 May the fleas of a thousand camels infest your most sensitive regions, OE wiki ! May 07 13:08:13 Laibsch: I'm pretty sure it does May 07 13:08:16 cool May 07 13:08:21 I'll give it a shot May 07 13:08:29 Laibsch: I was running into the same issue with gpe May 07 13:08:40 OK May 07 13:08:54 Laibsch: make sure "pointercal" is in your image (not sure what pulls it in) May 07 13:14:52 why is task-base-bluetooth and task-base-wifi 'hardcoded' in task-base and doesnt depend on DISTRO_FEATURES? May 07 13:19:02 Hi, does bitbake remove file by /bin/rm or remove in python ? May 07 13:19:47 * Julian_chu got a very very strange problem :-( May 07 13:20:24 Julian_chu: still deleting all yur directories? May 07 13:20:53 XorA: yeah.....4 times (crying) May 07 13:21:07 Julian_chu: bizarre May 07 13:21:12 XorA: but it never happens on other computers May 07 13:21:27 Julian_chu: check you .bashrc and .profile for wierd variables May 07 13:21:53 Julian_chu: normally this sort of thing comes down to aliases or variables with wierd values May 07 13:24:27 XorA: yeah, the .profile never changed since 2/12 May 07 13:25:18 you could do alias and export on machine where it is broken and where it works, then diff them May 07 13:25:23 XorA: I only add some alias in bashrc, likes bpi, bbb, qq,cdo, cdk .....:P May 07 13:25:54 jeremy_laine: Would MACHINE_EXTRA_RRECOMMENDS = "kernel-module-uvesafb v86d" be appropriate for quemarm, too? May 07 13:25:59 * Laibsch uses quemarm May 07 13:26:10 qemuarm, even May 07 13:27:28 Laibsch: no, not at all May 07 13:27:45 XorA: it always stop in the same step May 07 13:27:55 so, then jeremy's patch won't fix May 07 13:27:57 Julian_chu: what step? May 07 13:28:01 !oebug 3472 May 07 13:28:02 * * Bug 3472, Status: RESOLVED (WONTFIX), Created: 2007-12-13 10:08 May 07 13:28:03 * * : apache2-2.2.3-r5-do_configure May 07 13:28:04 * * http://bugs.openembedded.net/show_bug.cgi?id=3472 May 07 13:28:09 Sorry, wrong number May 07 13:28:17 !oebug 3742 May 07 13:28:18 * * Bug 3742, Status: NEW, Created: 2008-02-05 14:08 May 07 13:28:19 * * : Latest opie images fail calibration (in qemu) May 07 13:28:20 * * http://bugs.openembedded.net/show_bug.cgi?id=3742 May 07 13:28:35 RP, jeremy_laine: Is that the case? May 07 13:28:57 RP: Why is it not appropriate for qemuarm? May 07 13:29:03 Laibsch: Not sure about the MACHINE_FEATURES change May 07 13:29:20 Laibsch: because qemuarm doesn't use vesa. v86d is x86 only May 07 13:29:26 XorA: e-wm-0.16.999.042+cvs20080506-r5: task do_unpack: started May 07 13:29:36 It will not help in the slightest May 07 13:29:49 XorA: and it failed because the sources dir was deleted May 07 13:30:18 XorA: and there only remains 6 stamp files May 07 13:30:38 Julian_chu: that is bizarre May 07 13:30:43 XorA: yeah, ha May 07 13:30:46 Julian_chu: when did you last reboot that machine? May 07 13:30:48 RP: OK, I guess we need something similar, then, right? May 07 13:30:52 RP: what you need is the pointercal changes May 07 13:30:56 XorA: few days ago May 07 13:31:06 * XorA wonders about file system corruption May 07 13:31:18 RP: should work if you copy the qemux86/pointercal to qemuarm/pointercal May 07 13:31:22 Laibsch: The pointercal files sound more likely but calibration should work May 07 13:31:25 Julian_chu: are you one of the reiser people? May 07 13:31:29 XorA: and before the bizarre thing came, I could build meta-toolchain successful May 07 13:32:09 XorA: it only happens on my computer :P May 07 13:32:26 Julian_chu: thats why I wonder about fs corrution May 07 13:32:48 RP, Laibsch : the v86d addition makes the boot progress screen work for me May 07 13:32:58 thats the sort of wierd shit I used to see with reiser when it ate its own filesystem May 07 13:33:54 jeremy_laine: Yes, thats fine for x86 but it ***will not*** help qemuarm May 07 13:34:07 XorA: hmm.....probably May 07 13:34:21 RP: sure, for qemuarm you just want the pointercal, no? May 07 13:34:37 jeremy_laine: That could be the problem, yes May 07 13:34:47 jeremy_laine: but manual calibration should also work and if that doesn't there is some other issue May 07 13:35:32 I just want it clear that adding anything to do with vesa and v86d will not help qemuarm and anyone trying that is wasting their time May 07 13:36:06 XorA: if it about fs corrution, any suggestions ? May 07 13:36:17 Julian_chu: burn the computer May 07 13:36:48 RP: sure, that's why I only added v86d to qemux86.conf, not qemu.inc May 07 13:36:51 XorA: hahahaha, good idea XD May 07 13:37:18 jeremy_laine: I suspect you know this, I just want to make it clear to Laibsch too ;-) May 07 13:37:27 demonic posessed computers are a pain May 07 13:38:03 RP: understood May 07 13:38:10 manual calibration does not work May 07 13:38:52 Laibsch: You could try adding the pointercal file from poky and see if it helps May 07 13:39:07 XorA: reboot.....hopes the ghost will leave my computer May 07 13:39:09 If it does that will just mask another problem though... May 07 13:42:39 * Laibsch notes that the bug count for http://bugs.openembedded.net/buglist.cgi?keywords=patch&resolution=--- is only down by two since yesterday May 07 13:45:59 XorA: got roman ritual ? May 07 13:48:55 johncylee: I would try fire and chainsaw, fixes everything May 07 13:49:05 johncylee: or shotgun if its after yah brains May 07 13:49:26 ??? May 07 13:50:02 XorA: julian's pc will just attempt suicide right after bitbake e-wm May 07 13:50:16 * mwester prefers railgun May 07 13:50:24 johncylee: I saw that, I dont get it at all May 07 13:50:54 XorA: me neither May 07 13:51:09 XorA: and his bitbake will send "NOTE: xxxx failed xxxx" to bash! May 07 13:51:32 XorA: so bash will complain things like "command NOTE does not exist" May 07 13:51:41 werder and wierder May 07 13:51:49 * XorA suspects Tzeentch May 07 13:51:58 XorA: creepy May 07 13:52:20 XorA: Tzeentch <- what's that May 07 13:53:40 johncylee: http://en.wikipedia.org/wiki/Tzeentch May 07 14:11:05 * * OE Bug 4219 has been RESOLVED (FIXED) by May 07 14:11:07 * * FrobTADS May 07 14:11:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4219 May 07 14:44:04 bye May 07 14:54:22 ~seen bluelightning May 07 14:54:24 bluelightning was last seen on IRC in channel #opie, 8d 16h 55m 13s ago, saying: 'ok....'. May 07 14:58:04 * * OE Bug 4113 has been RESOLVED (REMIND) by May 07 14:58:06 * * Callweaver VOIP PBX package recipes and files May 07 14:58:08 * * http://bugs.openembedded.net/show_bug.cgi?id=4113 May 07 15:21:19 03crofton 07org.oe.dev * rda00e30f... 10/ (1 conf/machine/davinci-dvevm.conf): May 07 15:21:19 davinci-dvevm.conf : Update u-boot version to release and correct kernel May 07 15:21:19 entry point. May 07 15:21:23 03crofton 07org.oe.dev * r2b28a077... 10/ (3 files in 3 dirs): linux-davinci : Update to current git rev and update defconfig. May 07 15:25:29 hi May 07 16:22:21 RP: maybe you want to take a look at May 07 16:22:28 !oebug 2948 May 07 16:22:29 * * Bug 2948, Status: NEW, Created: 2007-09-06 23:56 May 07 16:22:30 * * m3l3m01t(AT)gmail.com: Packaged Staging feature May 07 16:22:31 * * http://bugs.openembedded.net/show_bug.cgi?id=2948 May 07 16:22:48 and see what is left to steal ;-) May 07 16:45:06 * * OE Bug 4227 has been created by bugs.openembedded.org(AT)rolf.leggewie.biz May 07 16:45:08 * * please backport fakeroot May 07 16:45:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4227 May 07 16:45:29 re May 07 16:50:55 hi woglinde May 07 17:06:37 hello, how does one uninstall a package from the staging directory? May 07 17:08:25 also, how does one build the depend package? I mean for instance if one builds a mediaplayer app which depends on ffmpeg, "bitbake mediaplayer" will build the ipk for mediaplayer but not for ffmpeg. Is there a way to tell bitbake to build both? May 07 17:26:56 Nahor: Are you talking about runtime dependencies not being built? May 07 17:27:21 the binaries are built just fine May 07 17:27:29 it's the packages that are not build May 07 17:28:11 hm? May 07 17:28:46 tell me what you do, what you expect as outcome and the outcome you see May 07 17:29:17 if I then do a "bitbake ffmpeg", it goes strait to the install/package tasks (I don't remember out of my head the exact task but for sure it skips the compile part) May 07 17:29:39 k, so "mediaplayer" depends on "ffmpeg" May 07 17:29:46 none are built yet May 07 17:29:53 I run "bitbake ffmpeg" May 07 17:30:02 bitbake compile ffmpeg and mediaplayer May 07 17:30:12 it also builds the ipk for mediaplayer May 07 17:30:21 but it doesn't build the ipk for ffmpeg May 07 17:30:39 Try May 07 17:30:39 BB_DEFAULT_TASK = "buildall" May 07 17:30:42 in local.conf May 07 17:30:51 although I still don't fully understand your problem May 07 17:31:15 But I believe it is solved by above or "bitbake -c buildall $target" May 07 17:31:26 I'd like bitbake to also build the package for ffmpeg, so that I can install mediaplyer and its dependencies on my device May 07 17:31:37 in this case "bitbake -c buildall mediaplayer May 07 17:31:38 " May 07 17:32:00 Nahor: I am pretty sure you want that line in local.conf May 07 17:32:05 and you'll be happy May 07 17:32:24 RP: Shouldn't we make that default for people to opt out? May 07 17:32:44 New people will run into trouble "only" for the benefit of saving CPU cycles May 07 17:32:58 People in the know could just opt out May 07 17:33:07 thanks Laibsch, I'll try that May 07 17:40:05 * * OE Bug 4228 has been created by bugs.openembedded.org(AT)rolf.leggewie.biz May 07 17:40:07 * * various likely targets for backporting May 07 17:40:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4228 May 07 17:43:19 RP: I assume that 3742 is caused by the non-existant pointercal file for qemuarm in .dev May 07 17:43:38 I'll compile test image and eventually push to .dev May 07 17:44:58 I think pushing the nokia pointercal at the same time makes sense May 07 17:50:06 what changed in udev May 07 17:50:18 * Crofton is scared by udev changes .... May 07 17:55:05 RP: Copying the pointercal file fixed the problem May 07 17:55:22 You said in that case that meant it was hiding the real root cause? May 07 17:55:28 another bitbake question: ffmpeg has two packages 0.4.9-pre1 and svn. How do I force it to use the svn one? May 07 17:56:11 If I use "PREFERRED_VERSION_ffmpeg = "svn"" in local.conf, it says: preferred version"svn" of ffmpeg not available (for item ffmpeg) May 07 17:56:35 then goes to compile 0.4.9-pre1 May 07 18:00:51 svn is not a version May 07 18:01:08 I guess the svn file has set DEFAULT_PREFERENCE May 07 18:01:16 then you'd comment out that line May 07 18:01:23 yes, it's set to "-1" May 07 18:01:42 Beware that there is a reason (usually) that it is not made the default May 07 18:02:10 It may not compile or segfault or whatever May 07 18:02:30 Look through "mtn log $path2svnbbfile" to find out more May 07 18:03:00 Laibsch, I have a memory of someone having an olsr problem with the gumstix fork May 07 18:04:38 I'm updating my copy of that data and will try bitbaking olsrd May 07 18:05:01 I think it's because ffmpeg doesn't have a version number any more (0.4.9 is the last one and fairly old) so svn has to point to the HEAD which can indeed be dangerous May 07 18:05:06 but thanks for the warning May 07 18:07:53 Nahor: OE could point to a known good svn revision May 07 18:08:00 so there has to be another reason May 07 18:08:10 but it could just be that nobody cared enough May 07 18:08:26 Let me know if you can successfully compile and run the current version May 07 18:08:31 I'll then "pin" it May 07 18:08:42 k May 07 18:11:17 Crofton: which distro do the gumstix people use normally? May 07 18:11:40 angstrom/oe from around October May 07 18:11:45 pre-dates the stable branch May 07 18:11:59 changes are kept in a collection so they are esy to find May 07 18:16:00 btw, .dev uses uclibc+svn, what features does svn have in addition? or in other words: what was the motivation behind going with an svn version rather than with the released one? May 07 18:24:00 Laibsch: the "pin" stuff, is it the "SRCREV_pn" in ./conf/distro/include/sane-srcrevs.inc? May 07 18:25:12 jineld there are many fixes May 07 18:25:16 in svn May 07 18:25:36 so we dont need to strip the check ins out May 07 18:25:39 as patches May 07 18:25:50 but I didnt change this May 07 18:25:55 ask cbrake or khem May 07 18:26:33 k.. well was just curious May 07 18:26:44 thought if it is worth switching/trying it May 07 18:26:57 btw do you know why the OE toolchain for uclibc is compiled with --disable-__cxa_atexit option? May 07 18:27:34 sorry cant rember please try search in the bugtracker May 07 18:27:43 I think there is the reason describe May 07 18:28:53 hm I wonder why I am the only one compiling uclibc and got errors with libiconv May 07 18:29:08 hmm, only one bug responds to that search, and thats about something different, just match in the log May 07 18:29:45 I didnot try libiconvin .dev lately, but uclibc + libiconv in stable work fine for me May 07 18:32:05 hm talking about dev May 07 18:32:14 zecke put the recipe in May 07 18:32:23 but the autotool stuff needs rewrite May 07 18:59:13 Nahor: yes May 07 18:59:46 Crofton: angstrom distro or what distro? May 07 19:00:22 gumstix? angstrom May 07 19:00:42 re May 07 19:00:50 re flo May 07 19:16:05 * * OE Bug 4229 has been created by nahor.j+openembedded(AT)gmail.com May 07 19:16:07 * * Wrong license for ffmpeg package. May 07 19:16:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4229 May 07 19:21:46 HEAD uclibc-initial-0.9.28 fails during install, using host /include and /lib paths. Anyone else runs into these? May 07 19:30:06 03Laibsch 07org.oe.dev * r53dccbcb... 10/ (5 files in 4 dirs): pointercal: backpush pointercal files for qemuarm and htcuniversal from poky. Closes 3742 (it at least masks the superficial problem). May 07 19:58:25 rp ping? May 07 20:06:41 hi all May 07 20:06:57 gm May 07 20:07:32 i have a trouble ... seem as udev don't create many entry on /dev ... which version of udev you suggest with a kernel 2.6.20.x ??? May 07 20:07:33 hi gremlin May 07 20:07:37 hi crofton May 07 20:07:42 hi Crofton woglinde May 07 20:14:17 03woglinde2 07org.oe.dev * rfea7dedc... 10/ (11 files in 3 dirs): May 07 20:14:17 gmp: update to version 4.2.2 May 07 20:14:17 * seems to work May 07 20:14:17 * add patches 02_noexec-stack.diff and 03_gnu89-inline.diff from debian May 07 20:14:17 * add mpf_set_str_c.diff from the gmp-website May 07 20:16:37 I have some newb questions... is do_rootfs the task when the actual deploy images are created? or is that do_deploy? May 07 20:16:56 basically i'm trying to make some manual modifications to the rootfs before it becomes a deployed image May 07 20:17:07 woglinde: pong May 07 20:17:08 and was wondering how you all do that May 07 20:20:29 Fique: yes, its the rootfs task May 07 20:20:52 Fique: image.bbclass and probably rootfs_ipk.bbclass May 07 20:21:07 Laibsch: I'd be ok adding that to the local.conf.sample May 07 20:21:09 thanks i'll take a look at doing things before then May 07 20:21:33 Laibsch: and I dealt with that packaged staging bug May 07 20:22:04 rp hm what is the sanest way if you have package.bb and package-bla.bb which only differs in configure option but should use same source and same patches May 07 20:22:07 * * OE Bug 2948 has been RESOLVED (WONTFIX) by rpurdie(AT)rpsys.net May 07 20:22:09 * * Packaged Staging feature May 07 20:22:11 * * http://bugs.openembedded.net/show_bug.cgi?id=2948 May 07 20:23:14 woglinde: package-bla.bb would "require package.bb" then change the option May 07 20:23:25 aeh rigth May 07 20:23:27 t May 07 20:23:29 args May 07 20:23:39 ~lart directfb author May 07 20:23:40 * ibot strangles directfb author with a doohicky mouse cord May 07 20:25:21 ups May 07 20:25:22 I meant May 07 20:25:37 ~lart xine-lib package wroter May 07 20:25:37 * ibot urinates on xine-lib package wroter May 07 20:26:11 hm lart both May 07 20:26:14 haha May 07 20:27:31 morning again May 07 20:27:33 * RP worries about woglinde May 07 20:27:43 morning hrw! May 07 20:27:46 re hrw May 07 20:27:50 more power under desk now May 07 20:28:06 hrw: You got your new space heater? :) May 07 20:28:51 RP: q6600, 4gb ram and mainboard a bit higher then your May 07 20:29:06 but same gf6600gt (but new thermal paste and cleaned fan) May 07 20:29:15 new 550w psu also May 07 20:29:27 ah.. and alix is now under desk too ;D May 07 20:30:04 but no cpufreq ;( May 07 20:31:09 hrw: nice :) May 07 20:31:20 hrw: I do appreciate mine for raw build power May 07 20:31:42 hrw: so you decided to go with Intel this time? May 07 20:31:50 cbrake: yes May 07 20:32:11 cbrake: few phenom contra quad benchmarks helped May 07 20:32:37 cbrake, thanks for the autofoo article, I've sent the link to few people I know May 07 20:32:43 ah. and 500g sata hdd May 07 20:33:05 * * OE Bug 4231 has been created by javiroman(AT)kernel-labs.org May 07 20:33:07 Crofton: your welcome -- glad its useful. I hope to expand it in the future. May 07 20:33:07 * * org.openembedded.stable: gconf-dbus_svn.bb do_fecth fails May 07 20:33:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4231 May 07 20:34:08 Crofton: I can't ever remember autofoo stuff from one encounter to the next, so I'm trying to create a reference of the stuff I use often :-) May 07 20:34:32 heh May 07 20:34:47 I know what you mean May 07 20:36:00 hrw: any idea how much HD performance affects builds OE? I'm running an older PCI 3-ware raid card here and wonder if it is worth upgrading. May 07 20:37:58 cbrake: no idea May 07 20:38:03 never checked too much May 07 20:38:17 cbrake: IMO bitbake is as IO bound as it is CPU bound May 07 20:38:19 time to enlarge bb threads May 07 20:38:49 and connect harddrive from alix to get sources ;( May 07 20:39:09 I need to buy cf card May 07 21:11:23 hrw: q6600 overclocks nicely; what motherboard do you have? May 07 21:14:42 mwester I have a nice box with 2 E5462 :) May 07 21:15:25 03polyonymous 07org.oe.dev * rf3aafac3... 10/ (1 site/common-glibc): May 07 21:15:25 pth: fix it for (at least) gentoo multilib system. May 07 21:15:25 actually, it's not a fix, but rather an attemt to mimick its behaviour on other May 07 21:15:25 systems, which is wrong, anyway, since it attempts checks against build system. May 07 21:15:25 But since it works, perhaps it should do for now. On the other hand, maybe opkg May 07 21:15:26 crashes somewhere around here, as gpgme depends on pth. May 07 21:15:30 BTW, long time no commits ;-) May 07 21:16:08 You know, before I became involved with OE, my main Linux system was a P-III, and it worked just fine. Now I have a quad-core with 4GB ram, and I'm trying to overclock it... May 07 21:16:26 OE is expensive! :D May 07 21:16:31 mwester and what was your system in '89? :) May 07 21:16:50 hehe! Abacus, I think. May 07 21:17:05 * * OE Bug 3988 has been RESOLVED (FIXED) by May 07 21:17:07 * * pth-2.0.2-r0 does not compile May 07 21:17:09 * * http://bugs.openembedded.net/show_bug.cgi?id=3988 May 07 21:17:20 I didn't have one of my own, but I was working with russian pdp-11 clones back then ;-) May 07 21:17:33 And, I believe, I've seen 286 too! May 07 21:17:51 polyonymous you are from GDr? May 07 21:18:04 woglinde, nope, from russia. May 07 21:18:19 polyonymous hehe okay May 07 21:18:32 So I was close to the nest of those clonse ;-) May 07 21:18:35 clones too :) May 07 21:18:42 we have one pdp-11 clone at the university May 07 21:18:44 mwester: gigabyte ep35c-ds3r May 07 21:18:54 and pdp-11 next to ot May 07 21:19:06 but both rotten in the entry hall May 07 21:19:08 woglinde, we have pdp-11 emulator in OE :) May 07 21:19:09 *g* May 07 21:19:31 * flo_lap wants a real one May 07 21:19:36 (because I couldn't resist temptation having my first love in my pocket) May 07 21:19:45 fko pdp-11? May 07 21:20:00 flo to build a refrigerator in? May 07 21:20:08 I don't want one. I want to live with my sweet memories. May 07 21:21:18 I want HP2623A terminal May 07 21:21:37 I want tea. brb. May 07 21:23:22 woglinde: no... I like ancient technology :) May 07 21:26:49 450G empty ext3 partition has 199M of used space.. May 07 21:27:18 http://tv.timbormans.com/ - last.fm + youtube in one May 07 21:28:35 flo_lap, ancient! Are you calling me an old fart?! :) May 07 21:29:08 polyonymous_: how old you are btw? May 07 21:29:29 well, yes... I *am* an old fart... 35 May 07 21:29:42 polyonymous_: hehe May 07 21:29:48 35 and played with pdp11? May 07 21:30:02 I am 32 May 07 21:30:19 hrw, I think you should know that russia was way behind in technology May 07 21:30:31 hrw: hum why not..? I play with a lot of old toys :) May 07 21:30:31 89 May 07 21:30:34 polyonymous_: I know, I am from same side of Iron Curtain May 07 21:30:44 * woglinde calculates May 07 21:30:46 hrw, that's why I say you should know that. May 07 21:31:13 polyonymous_: can you get some Pentagram or Scorpion machines? May 07 21:31:23 And I started with EC1840 (russian ibm clone, I believe, with punchcards and everything) at the age of 12 or so :) May 07 21:31:38 uh May 07 21:31:39 hrw, What are those? May 07 21:31:51 first computer was atari 65xe for me May 07 21:31:58 polyonymous_: russian zx spectrum clones May 07 21:32:39 hrw, Are you sure they are named like that? I have never heard about them. May 07 21:33:10 or Pentagon and Scorpio - have to look May 07 21:33:40 Still doesn't ring a bell... May 07 21:34:30 Scorpion ZS-256 and Pentagon-48 May 07 21:34:51 also Pentagon-128, Moskva-128, Moskva May 07 21:35:54 http://pentagon.nedopc.com/ May 07 21:36:01 And when were these manufactured? May 07 21:36:05 aha May 07 21:37:14 polyonymous_: 1988 mostly May 07 21:37:31 http://www.nvg.ntnu.no/sinclair/computers/clones/russian.htm May 07 21:37:46 hrw, I'm really surprised there was something called pentagon back then :) May 07 21:39:01 hrw, well, I have to admit I never heard of them. Back then I was deep in the core of pdp-11 which I used to programm using switches on the front panel :) May 07 21:39:57 I haven't seen most of the clones on the page... May 07 21:59:04 * * OE Bug 4232 has been created by spam(AT)campana.vi.it May 07 21:59:07 * * Impossible to build linux-hackndev-2.6 May 07 21:59:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4232 May 07 22:02:18 bye May 07 22:02:25 nite hrw May 07 22:09:27 argh. push takes ages. May 07 22:14:05 * * OE Bug 4233 has been created by spam(AT)campana.vi.it May 07 22:14:07 * * trivial bug in org.openembedded.documentation May 07 22:14:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4233 May 08 02:11:36 03Laibsch 07org.oe.dev * rb8e46b79... 10/ (14 files in 7 dirs): gmp: consolidate common patches unchanged across versions into files/ May 08 02:36:51 hello. how can i pass a username to the svn fetcher in my recipie? May 08 02:37:45 I tried setting FETCHCOMMANDsvn="checkout --username ..." May 08 02:38:37 didn't make any difference **** ENDING LOGGING AT Thu May 08 02:59:57 2008