**** BEGIN LOGGING AT Mon Oct 14 02:59:58 2013 Oct 14 07:53:15 JaMa: thanks for qt5.1 fontconfig/freetype, will get some tests today. btw, our autobuilder caught the issue on Friday about "fatal error: freetype/config/ftheader.h: No such file or directory". good to see that you fixed that too ;-) Oct 14 08:15:19 good morning Oct 14 08:16:17 hi mckoan Oct 14 08:17:04 hi there Oct 14 08:17:29 is there any chance that Andrea Adami is here? and does anyone knows his nickname? Oct 14 08:17:48 ant_work Oct 14 08:18:01 hi Oct 14 08:18:10 hi Oct 14 08:19:01 ant_work: I'm Vasily Khoruzhick, you wrote me a mail about h1940 kernel Oct 14 08:19:11 anarsoul: as I've written that h1940 is now being coverted to DT Oct 14 08:19:30 anarsoul: it would be the first one in meta-handheld Oct 14 08:19:36 but I'm pretty sure that DT support for s3c2410 is not ready Oct 14 08:19:49 (yet) Oct 14 08:19:58 or you're working on it? :) Oct 14 08:20:34 nope, but the SOC guys are active Oct 14 08:20:40 hehe Oct 14 08:21:07 LinusW is now trying to unbork sa1100 Oct 14 08:21:29 Heiko Stübner is working on s3c2416, but I know no one who works on s3c2410 and s3c2440/2442 Oct 14 08:22:42 ok, if it is possible then pls refresh the .config and provide a shrinked defconfig for 3.10 or 3.12-rc Oct 14 08:22:51 old-style Oct 14 08:22:54 (and actually there's no DT-capable bootloader for h1940, at least for now) Oct 14 08:23:47 I'm not sure if haret supports DT, and my u-boot port is quite old, so no DT for sure Oct 14 08:24:04 ant_work: ok, will try tonight Oct 14 08:27:13 thx Oct 14 08:38:33 morning all Oct 14 08:39:02 gi silvio Oct 14 08:39:37 woglinde: I've updated even for your old simpad ;) Oct 14 08:40:20 woglinde: if you had any free time, pls send a defconfig for 3.10 - 3.12, thx Oct 14 08:42:02 ant_work yes I saw the patches Oct 14 08:42:09 but I have no time to play with it Oct 14 08:53:38 anarsoul: so used an appended DT for the non-DT aware bootloaders Oct 14 08:54:02 koen: yeah, but a lot of drivers for h1940 need convertion to DT Oct 14 08:54:07 especially display part Oct 14 08:56:34 you can do hybrid boots Oct 14 08:56:46 enable the h1940 boardfile and enable DT support Oct 14 08:57:02 it's how most ARM DT boards boot nowadays Oct 14 08:57:16 ah, ok Oct 14 08:57:17 there are *very* few boards that can do a 100% DT boot and work properly Oct 14 08:57:44 it would be nice to unify s3c24xx ipaqs anyway... Oct 14 08:57:53 rx1950, rx3xxx and h1940 Oct 14 08:58:55 anarsoul: note that at the beginning for linux-yocto we'll use the defconfig which will be re-expanded with make oldconfig Oct 14 08:59:06 this is the 'compat-mode' Oct 14 08:59:51 ant_work: I didn't use OE for a while, so I'll make kernel working first with simple buildroot-based initramfs and then will try to switch to OE Oct 14 09:00:08 anarsoul: yes, just do make savedefconfig pls Oct 14 09:00:31 I'll add to current linux-yocto(-dev) Oct 14 09:02:34 I think it would be nice to add s3c24xx support into qemu, so oe devs will be able to test h1940 stuff :) Oct 14 09:02:49 anarsoul: does that ipaq have a rotated framebufer? Seems not needing fbcon=rotate:1 Oct 14 09:03:00 ant_work: it doesn't Oct 14 09:03:06 ok Oct 14 09:04:06 heh, need to patch formfactor Oct 14 09:11:46 morning all Oct 14 09:12:22 anarsoul: hey, long time no see :) Oct 14 09:12:33 hi bluelightning :) Oct 14 09:23:51 hi bluelightning and stefan_schmidt Oct 14 09:43:19 hi woglinde Oct 14 09:47:48 hi bluelightning , hi woglinde Oct 14 10:04:15 ndec: yes it's probably the reason why I've enabled fontconfig by default :) Oct 14 10:04:33 well, now you know, at least@ Oct 14 10:04:35 ! Oct 14 11:09:16 Hi! Does anyone have an idea why this image https://github.com/Angstrom-distribution/meta-ti/blob/master/recipes-misc/images/cloud9-image.bb brings up the USB storage (g_multi module) while the one it is based on https://github.com/Angstrom-distribution/meta-ti/blob/master/recipes-misc/images/ti-hw-bringup-image.bb does not load the USB storage? What might be the package that makes the difference? Oct 14 13:13:38 bluelightning, Did I need to do anything else in addition to mailing my patches to the mailing list for bringing in sthttpd and adding the systemd recipe? Oct 14 13:14:28 exosyst: well, the patch series is a little odd because it mixes a backport and an add of a feature Oct 14 13:14:48 exosyst: ideally these should be separate Oct 14 13:15:22 bluelightning, Well they are standalone but ideally work better together. One to provide the backport, one brings in sysd Oct 14 13:16:47 I can apply the systemd patch today, but the backport really needs to be a separate patch marked with [dylan] in the subject Oct 14 13:17:28 bluelightning, OK I'll do that, same list? Oct 14 13:18:27 exosyst: yep, but it's probably best if I apply this patch first and then you'd just be adding the fully up-to-date version to dylan after that Oct 14 13:19:16 bluelightning, OK, if that works for you Oct 14 13:20:53 exosyst: just noticed, the RPROVIDES/RREPLACES/RCONFLICTS lines shouldn't be needed Oct 14 13:21:34 these are only there in other recipes for compatibility with the old meta-systemd Oct 14 13:21:53 where corresponding bbappends existed in that layer Oct 14 13:22:09 I can just tweak the commit if that's OK with you Oct 14 13:22:39 bluelightning, that's fine. First time trying to help with OE so if there's anything I can do let me know -- just might require some handholding Oct 14 13:22:42 ah, there might be another problem as well Oct 14 13:23:29 I'm not sure, but I think systemd in compatibility mode will try to match up the initscript and service name; you've named the service sthttpd and the initscript is called thttpd Oct 14 13:23:51 I'm thinking that because they're named differently, systemd might try to start it tice Oct 14 13:23:54 *twice Oct 14 13:25:31 I assume wwe can use recipes-core in our own layers, right? Oct 14 13:25:53 bluelightning, Hmm - I did email the maintainer of the project to see if there was an official name as it's listed as just Supported thttpd on the site rather than just sthttpd Oct 14 13:26:55 Crofton|work: sure Oct 14 13:27:32 I thought so, but someone asked me Oct 14 13:27:44 and reminded me I should use it :) Oct 14 13:28:05 bluelightning, I hadn't noticed the issue but can redo it - and resend just the systemd fix? Oct 14 13:28:09 Crofton|work: ideally people wouldn't have stuff in there that wasn't bbappends/replacements for recipes in OE-Core, but there's no restriction Oct 14 13:28:20 well, for meta-sdr Oct 14 13:28:31 we have some packages that are core Oct 14 13:28:42 bluelightning, Actually I need the backport done first for that to work Oct 14 13:28:57 but not so core they belong in oe-core :) Oct 14 13:29:12 Crofton|work: well I'll leave it up to your judgement :) Oct 14 13:29:29 I was going to use another name, but someones hair stood on end :) Oct 14 13:29:52 exosyst: can you not just adjust the commit you already sent on your branch and just send that one alone? Oct 14 13:30:07 s/adjust/amend/ Oct 14 13:30:17 to use official terminology Oct 14 13:32:51 I've done a soft git reset. I was going to redo it all again and rename the service to be thttpd.service - that no good? Oct 14 13:33:26 right, that'll work in the same way yes Oct 14 13:34:25 bluelightning, Cool - I've removed the RPROVIDES, REPLACES, CONFLICTS as well Oct 14 13:35:21 Can I still commit as before or should I do two, seperate commits - one as [dylan] backport and the the next as [meta-webserver] bluelightning? Oct 14 13:37:13 exosyst: they'd both be marked [meta-webserver] Oct 14 13:37:22 exosyst: as long as the patch applies on top of master let's just concentrate on that for now Oct 14 13:57:52 bluelightning, So that's the seperate dylan mentioning one. Oct 14 14:12:46 OK, I think that solves it - two seperate commits and hopefully it's correct this time - let me know if I can redo anything bluelightning - my Git-fu could use some more practive Oct 14 14:17:35 exosyst: so typically we would mark what you've just sent as "PATCH v2" and include the comments about what changed underneath the --, or a cover note (reason being, notes about what changed from the previous patch aren't particularly meaningful when looking back at the git history later) Oct 14 14:20:39 bluelightning, right - yeah, I may have been a bit hasty and had seen a few v2 messages on the list (only been a member for a few days) - Maybe i can update the wiki with those notes too. Oct 14 14:34:07 Did you want me to re-resend or can it be accepted in wonky form? Oct 14 14:39:03 it's OK, I can tweak it Oct 14 14:57:12 exosyst: actually I just noticed there are hardcoded paths in the service file and we are taking care of those for the initscript; I'll send out a version with the fixes in it Oct 14 14:58:24 bluelightning, that's great. Hopefully I can learn the lessons for next time (that's right, I want to continue "helping" - you poor bastards) Oct 14 15:02:03 err... hmm Oct 14 15:02:14 just got a checksum failure trying to build-test it :( Oct 14 15:04:40 oh, internal firewall fail, nm Oct 14 15:14:06 can someone please tell me what the flex-dev package is for? Oct 14 15:15:09 for some strange reason it is needed in order to load the g_multi USB gadget module on BBB... Oct 14 15:25:47 Anguel_: that sounds very unlikely - flex-dev would contain the development headers and other related files for the GNU flex lexer Oct 14 15:26:05 not something a kernel module would ever itself use at runtime Oct 14 15:27:12 bluelightning: maybe some script? I just spent a lot of time and tracked it down to that package Oct 14 15:28:09 bluelightning: this script loads g_multi on BBB: https://github.com/beagleboard/meta-beagleboard/blob/master/meta-beagleboard-extras/recipes-support/usb-gadget/gadget-init/g-ether-load.sh Oct 14 15:28:58 if I build my image without flex-dev it does not auto-load the USB gadget, if I build it with it, it loads fine on startup Oct 14 15:31:58 Anguel_: that has to be a red herring Oct 14 15:32:06 something else is at work there Oct 14 15:35:35 bluelightning: sorry, I am not an expert, is it possible that this flex-dev actually installs something else? Oct 14 15:36:06 you can look at the contents of it fairly easily Oct 14 15:38:05 bluelightning: I did a search in the /sources/ for flex-dev files but found nothing Oct 14 15:39:11 flex-dev is a package produced by the "flex" recipe Oct 14 15:40:20 a ok Oct 14 15:40:28 Anguel_: I think it's much more likely that as a result of adding flex-dev you indirectly pulled in another package that caused the change in behaviour Oct 14 15:40:54 bluelightning: yes, I also suspected something like this Oct 14 15:42:52 exosyst: ok, patch v3 sent Oct 14 15:43:31 bluelightning: I see a flex.inc file that contains "inherit autotools gettext" -> is this a hint? Oct 14 15:43:52 no Oct 14 15:44:11 :( Oct 14 15:44:12 if you want to compare the images, you can use buildhistory to do that Oct 14 15:44:43 Anguel_: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#maintaining-build-output-quality Oct 14 15:45:04 see specifically 2.4.2.3. Using Build History to Gather Image Information Only Oct 14 15:45:18 then do a build of the image without flex-dev and with it Oct 14 15:45:35 well Oct 14 15:45:45 Anguel_: try running that script manually without flex-dev and seeing what happens Oct 14 15:46:11 use BUILDHISTORY_COMMIT = "1" so you can actually compare easily Oct 14 15:46:18 fwiw, flex-dev just contains a single header Oct 14 15:46:18 bluelightning, rburton: thanks, that is great help, I will check this Oct 14 15:46:39 actually this is probably easy Oct 14 15:46:48 that script uses "devmem2" Oct 14 15:46:52 I suspect that's what you're missing Oct 14 15:48:54 Anguel_: I'd suggest filing an "issue" in the meta-beagleboard github issue tracker mentioning that gadget-init needs an RDEPENDS on devmem2 to work Oct 14 15:49:57 bluelightning: can I just add devmem2 to the recipe directly? Oct 14 15:50:02 to test Oct 14 15:50:04 sure Oct 14 15:50:37 RDEPENDS_${PN} = "devmem2" Oct 14 15:51:20 ok, i now just added it to my image istead of flex_dev Oct 14 15:51:28 now bitbaking :) Oct 14 16:01:01 bluelightning: hmm, it does not load :( Oct 14 16:01:37 Anguel_: ok, so what happens when you do as rburton suggested i.e. run the script directly? Oct 14 16:02:03 bluelightning: -sh: /usr/bin/g-ether-load.sh: line 3: syntax error: unexpected "(" Oct 14 16:02:17 maybe the script has bashisms... Oct 14 16:02:32 yes Oct 14 16:02:40 () in function definitions is a bashism Oct 14 16:02:47 but the script is a #! /bin/sh Oct 14 16:03:07 but why does it work with flex-dev? :) Oct 14 16:03:28 Anguel_: probably because that results in bash being installed Oct 14 16:03:33 Anguel_: because flex-dev pulls in flex pulls in something else ... -> bash Oct 14 16:03:47 the script should be using functionname() {} Oct 14 16:03:52 without the function prefix, with curlies Oct 14 16:04:08 right, another thing that needs patching in meta-beagleboard Oct 14 16:07:56 rburton: so I shall just delete the "function" parts and retry? Oct 14 16:08:38 Anguel_: yes Oct 14 16:09:18 the "easy" fix is to make it #! /bin/bash and install bash, but that's distinctly worse. Oct 14 16:09:29 you might end up chasing bashisms all down that file though Oct 14 16:09:59 rburton: it seems to work now :) Oct 14 16:11:09 ok, I get "udhcpd: can't open '/var/lib/misc/udhcpd.leases': No such file or directory" but that is probably another thing Oct 14 16:11:29 and I don't get back to the prompt... Oct 14 16:13:16 udhcpd is probably waiting for a response Oct 14 16:13:22 just control-c it Oct 14 16:14:58 rburton: many thanks! shall I report that where bluelightning told me? Oct 14 16:15:22 yes Oct 14 16:16:02 omg, will do that tomorrow, have to go now :( thank you once again! Oct 14 17:22:08 bluelightning1, Just saw - thanks for the rework. I get a D-, must try harder :D Oct 14 17:24:32 exosyst: no problem :) Oct 14 17:28:29 exosyst will be come with the time and reading mailinglist Oct 14 18:16:31 Hello Oct 14 18:24:16 JaMa, excuse me. I'm looking at qt5 layer. Is there a custom sdk/meta-toolchain-qt5/whatever ? Oct 14 18:47:13 woglinde: you did not take the ppc patch yet for meta-java Oct 14 18:47:19 any issues you see with it ? Oct 14 18:47:35 khem sorry no time Oct 14 18:47:51 tomorrow I will look at it again Oct 14 18:57:12 no probs Oct 14 22:22:01 lumag: no, there isn't yet, few people said that they are working on it, but nobody sent patches yet **** ENDING LOGGING AT Tue Oct 15 02:59:58 2013