**** BEGIN LOGGING AT Tue Jul 09 03:00:00 2013 Jul 09 07:24:18 good morning Jul 09 08:05:21 can I use preffered version to use gcc 4.7.2 instead of 4.8.1 ? Jul 09 08:19:28 seems like I can, atleast it appears to be working Jul 09 08:20:10 erbo: it should work Jul 09 08:49:52 Hey, I have a question. I am trying to bitbake a package, and this package depends on another package. However... it seems as if the second package gets compiled for x86_64, while the project as a whole is ARM. This causes the first (dependent) package to fail, with an error that it ELFCLASS32 is (obviously) not right for building an ARM program. This is the error output: http://pastebin.com/727V2g9r Jul 09 08:50:06 It seems -c cleanall on perl or the original package does not help either. Anyone have any suggestions? Jul 09 09:01:42 bluelightning, I know it's bad custom to repeat oneself, but you've been helpful before, so... I have a question. I am trying to bitbake a package, and this package depends on another package. However... it seems as if the second package gets compiled for x86_64, while the project as a whole is ARM. This causes the first (dependent) package to fail, with an error that it ELFCLASS32 is (obviously) not right for building an ARM program. This is the Jul 09 09:01:42 error output: http://pastebin.com/727V2g9r. It seems -c cleanall on perl or the original package does not help either. Anyone have any suggestions? Jul 09 09:02:10 bluelightning, I did just post that here (before you joined) But I'm hoping you could give me a clue as to whats going on. Jul 09 09:05:05 Stygia: hi... I'm not exactly sure what's going on there I have to admit Jul 09 09:05:34 Stygia: is it saying that perl-data-dumper has been built for the wrong architecture? Jul 09 09:05:37 bluelightning, Alright... For some reason, it seems that my XSLoader (Perl) is trying to load the ARM .so file (static library, I think?) from the x86_64 pm.... Jul 09 09:05:53 good morning all Jul 09 09:05:54 bluelightning, I'm not quite sure. I'm cleaning it, and trying to compile it. I've named it cpan-data-dumper BTW. Jul 09 09:06:16 bluelightning, We don't really want/need x64 at all... It's for our boxes. Jul 09 09:06:33 Stygia: sure, understood Jul 09 09:06:40 bluelightning, So yea, unfortunately I'm sorta confused as to why it would even compile anything for x64. Jul 09 09:07:04 Stygia: maybe it's just using the wrong compiler... Jul 09 09:07:28 Stygia, it seems perl-native tries to load the wrong perl module, the one built for target arch Jul 09 09:07:31 bluelightning, It's... possible I suppose. But we haven't had trouble so far, or with other packages. Can you suggest some possible venues for me to explore? Jul 09 09:08:01 blitz00, Well it is supposed to build everything for the target arch. What I'm actually confused about is why it would try to include the x86_64 version at any point? Jul 09 09:08:30 Stygia: I'd be looking at the output of the configure step, see what it reports for the compiler / options Jul 09 09:09:19 bluelightning, What I send is essentially all of the output that's "off", but I didn't add -D Jul 09 09:09:30 bluelightning, Adding just one -D is the best in terms of readability, yea? Jul 09 09:10:51 Stygia: I usually just look at the task logs directly, i.e. temp/log.do_* under the workdir for the recipe Jul 09 09:13:58 oh, huge netsplit Jul 09 09:16:35 bluelightning, Hmm. It looks like I only have cleanstate, compile, configure, fetch, patch, populate and unpack as log.do_ entries for the failing package. Jul 09 09:17:00 Stygia: sure, it's configure I was recommending looking at Jul 09 09:18:11 bluelightning, Hmm. Well this is all of it: http://pastebin.com/DsQb23uR Jul 09 09:18:49 bluelightning, Note that it mentions the Data::Dumper module, but not in a "weird" way - it mentions the one from imx6sabresd which is the ARM system. Jul 09 09:19:21 bluelightning, But I suppose it is kinda odd that it also lists modules from x86_64 Jul 09 09:35:41 morning all Jul 09 09:38:44 morning pb_ Jul 09 09:51:11 hi pb_ Jul 09 09:51:17 hi bluelightning Jul 09 09:53:19 pb_: I'll have to ask you soon about a strange (for me) issue with CC and LD Jul 09 09:55:23 I have been told that particular code cannot be compiled with klcc because it's 'bootloader'- not unix code..?? gcc has no issues with it. Jul 09 09:57:50 it is a -Wl,-e,_start linking option Jul 09 10:02:40 morning #oe Jul 09 10:06:47 If I build the kernel for an imx6 board from meta-fsl-arm using GCC 4.8.1 it crashed during boot, but same sources built with 4.7.2 boots fine. Does anyone know if other hw-layers has seen something similar? Jul 09 10:07:46 erbo: I thought that issue had been fixed, are you using the latest updated 4.8.1 recipes? Jul 09 10:08:02 erbo: also I'm not sure but there may be a kernel patch needed as well Jul 09 10:08:37 ant_work: I don't recall exactly what the klcc wrapper does, but I wouldn't have thought it would matter very much. Jul 09 10:08:48 bluelightning: I did a pull today so I think I have all the latest Jul 09 10:09:06 but, by the same token, if this is bootloader code then presumably it doesn't need a C library at all so there might be no need to use klcc in the first place. Jul 09 10:09:35 otavio: any ideas to help erbo? ^ Jul 09 10:10:16 pb_: http://goo.gl/y7hK8 Jul 09 10:10:23 bluelightning: I've sent mails to the meta-freescale list so I guess otavio will see it there, isn't he usually asleep at this hour? Jul 09 10:10:57 pb_: I oscillate between silent failure and 'no -start' eror Jul 09 10:11:07 erbo: probably but hopefully he will see them when he comes online Jul 09 10:12:09 pb_: it's a single object part of kexec-tools and the previous versions did compile Jul 09 10:13:52 erbo: yes; none of my systems (pxa168, imx6, mx28) boot with gcc 4.8 :( Jul 09 10:15:13 apropos gcc 4.8... are there plans to make dylan buildable on gcc 4.8 systems (Fedora 19)? Jul 09 10:15:20 ant_work: oh, right. It's hard to say much about that without inspecting the code itself more carefully, I'm afraid. Jul 09 10:17:43 mmh... it's not a gcc problem with dylan but a something with pod2man Jul 09 10:18:04 pb_: np, thx Jul 09 10:19:59 ensc|w: the host gcc 4.8 fix is in dylan Jul 09 10:20:25 ensc|w: re pod2man do you mean the openssl failure? or something else? Jul 09 10:21:18 bluelightning: yes; the openssl failure and I see that it has been fixed recently Jul 09 10:21:47 ensc|w: right... I have that fix queued for backporting into dylan (on the paule/dylan-next branch) Jul 09 10:21:59 should be merged in the next few days Jul 09 10:23:06 no; it's already there ;) Jul 09 10:24:20 pb_: I've understood I have to use LD and not CC anyway. I'll simplify the flags to the bare minimum and readd them step by step Jul 09 10:25:15 ensc|w: ah right... Richard must have just done it this morning Jul 09 10:26:04 bluelightning: btw pls don't forget #53041 Jul 09 10:28:45 ant_work: ok, merged Jul 09 10:56:15 bluelightning: thx Jul 09 13:14:16 bluelightning: Hi Jul 09 13:14:19 erbo: Hi Jul 09 13:14:30 otavio: hi Jul 09 13:14:30 erbo: I think Gary did answer you already Jul 09 13:14:58 otavio: yes we exchanged a few mail on the list Jul 09 13:15:11 erbo: Can you prepare an e-mail with the fixes? Jul 09 13:15:27 erbo: even better if you send a patch ;-) Jul 09 13:16:04 otavio: sure, for meta-fsl-arm-extra or linux-boundary kernel? Jul 09 13:17:13 erbo: I'd say for meta-fsl-arm-extra so we can apply it soon and I talk to Eric to apply it in their tree. We will also need to apply it in imx tree for sure. Jul 09 14:22:32 erbo: Eric did push these changes to linux-boundary Jul 09 14:22:46 erbo: can you bump linux-boundary revision and test? Jul 09 14:22:58 erbo: if it works, please prepare a patch and send it to us Jul 09 14:23:18 otavio: sure, I saw his e-mail just as I pressed enter to send my patch for meta-fsl-arm-extra :) Jul 09 14:23:28 perfect timing :) Jul 09 14:28:34 erbo: good; thx Jul 09 14:43:12 anyone know of a vim syntax file for bitbake files that doesn't choke on functions / variables that end with _${PN}? Jul 09 14:53:39 otavio: patch sent (twice actually since I saw your email about signed-off too late) Jul 09 15:08:45 staticshock: I use the openembedded plugin without issues Jul 09 15:09:35 staticshock: http://git.openembedded.org/bitbake/log/contrib/vim Jul 09 15:09:45 hm, i think i've tried that one Jul 09 15:09:50 let me try again Jul 09 15:10:45 erbo: tested? Jul 09 15:10:46 it'd be nice if that was broken out into a separate plugin, so it was pathogen-friendly Jul 09 15:11:30 staticshock: https://github.com/kergoth/vim-bitbake Jul 09 15:11:35 not official, but it's what i use Jul 09 15:12:23 was split out with git filter-branch, so has history Jul 09 15:12:31 kergoth: yes, that's what i use Jul 09 15:12:45 it trips up on function names that end with _${PN} Jul 09 15:12:51 kergoth: it gives some update since dec.2012 Jul 09 15:13:10 ls Jul 09 15:13:12 oops Jul 09 15:13:22 kergoth: its only one commit for bbclass etc Jul 09 15:13:58 no idea what you're referring to. yes, upstream has had a couple commits since i created the repo, i implemented those same changes in separate commits Jul 09 15:14:07 staticshock: patches welcome :) Jul 09 15:14:24 staticshock: likely just need to allow $, {, and } in variable names in the syntax file Jul 09 15:14:50 i hacked at it for a bit, but i don't understand vim's syntax-describing-syntax well enough Jul 09 15:14:52 kergoth: I mean this: http://git.openembedded.org/bitbake/log/contrib/vim Jul 09 15:14:52 heh, that was the first vim syntax file i ever created, it could probably be done in a better way Jul 09 15:15:00 silviof: what about it? Jul 09 15:15:50 staticshock: heh, vim's language isn't always particularly pleasant, and syntax files can be a bit complicated at timesm :) Jul 09 15:15:59 staticshock: some days i wish it just used lua Jul 09 15:16:12 kergoth: sorry i have think you have do a "git filter-branch" on this repo Jul 09 15:16:15 i'll take a look at fixing _${PN} if i find the time today Jul 09 15:16:30 i have a hacky fix Jul 09 15:16:38 silviof: I still don't understand what you're talking about. again, I did a git filter-branch on bitbake to extract vim-bitbake, but the two repositories have deviated from one another since then Jul 09 15:16:53 i have a dotfiles repo that uses your repo as a submodule, kergoth, and applies my patch on top of it Jul 09 15:16:59 silviof: it's not a mirror, updating a filter-branched repository is nontrivial Jul 09 15:16:59 one sec, i'll pastebin the patch Jul 09 15:17:03 staticshock: cool, thanks Jul 09 15:17:17 not sure it's the ultimate solution, but it's as far as i got Jul 09 15:17:30 staticshock: honestly i expect everything we all do to that syntax file is basically a hack anyway, it's all voodoo :) Jul 09 15:17:45 it got a bit ugly due to the oddities of our format Jul 09 15:17:56 parsing bitbake files by anything is a bit of a mess Jul 09 15:18:33 kergoth: http://pastebin.ca/2422726 Jul 09 15:18:38 def'd functions have to terminate specially due to the indentation based format like python, function blocks shouldn't parse their contents, but should instead be parsed linewise until the terminating }, which basically means lexer states Jul 09 15:19:04 that fixes *some* of the problems i was seeing, but definitely not all Jul 09 15:19:18 heh, okay, might be a starting point Jul 09 15:19:21 and there were definitely some problems i just didn't get to see because i don't deal with files that had the relevent syntax Jul 09 15:19:33 the thing is, certain patterns are repeated in multiple places in the syntax file Jul 09 15:19:47 both the outer encompassing thing and the inner thing it includes tend to include parts of the same content Jul 09 15:19:54 due to not wanting to be greedy with the matches Jul 09 15:19:55 yeah, i saw Jul 09 15:20:25 yeah, i think all this patch does is add ${} to that one character group Jul 09 15:21:01 (verified: that's all it does) Jul 09 15:49:32 mmh... gcc 4.8 kernel problems seem to be buried deep within slub memory allocation code :( Jul 09 15:53:54 http://pastebin.com/79TQbtch Jul 09 16:17:08 kergoth: your git-origin thing looks pretty interesting. i'm gonna check that out. Jul 09 16:20:05 bluelightning1, I added you and Ross to the always accept bin for meta-xilnx Jul 09 16:20:24 Crofton|work: right, sorry about the x-post Jul 09 16:20:31 no worries Jul 09 16:28:22 staticshock: I actually don't remember if it worked, it's been years since i touched it, but i do think the concept is sound, something is clearly needed, maintaining long lived branches, e.g. stable branches, is rather painful without some form of cherry pick tracking Jul 09 16:28:26 heh Jul 09 16:29:17 boost 1.54 fails to builds while boost 1.53 builds just fine Jul 09 16:29:29 kergoth: yeah, i periodically use `git cherry` and then have to do a bunch of extra work just to wrangle with the output Jul 09 16:29:51 staticshock: https://github.com/pv/git-cherry-tree is the best tool i'ev seen for that so far Jul 09 16:30:00 staticshock: far superior to using cherry directly Jul 09 16:30:15 kergoth: thanks, i'll check it out Jul 09 16:30:24 staticshock: also checks to see if the patches are seen as already applied, has an interactive mode not unlike rebase -i for actual picking of the commits, etc Jul 09 16:30:41 cool Jul 09 16:30:45 downside is it pretty much has the same limitations as cherry, if the patch-id no longer matches due to con flicts, it won't know it's already applied Jul 09 16:30:52 but it's better than nothing Jul 09 16:32:00 * kergoth brushes up on vim syntax files Jul 09 16:35:10 * pb_ hands kergoth a copy of emacs Jul 09 16:56:47 moin Jul 09 16:57:22 field rials of rpi-mpd yocto/master build going well... Jul 09 17:34:53 kergoth: i'd just wish there was a --format flag for git-cherry Jul 09 17:35:27 because i'd like to see it formatted more like git-log --oneline Jul 09 17:37:14 git-cherry-tree outputs it in that format, or similar Jul 09 17:41:06 where would i put a packaging-time assertion in a bitbake recipe? Jul 09 17:41:36 do_configure? Jul 09 19:15:02 I'm trying to bbappend a recipe that has a bad "pkg_postinst_${PN}" Jul 09 19:15:20 I tried to do pkg_postinst_${PN} { \n \n} Jul 09 19:15:26 but I'm getting a Parse error Jul 09 19:15:35 it doesn't appear to like pkg_postinst to be overriden Jul 09 19:15:39 any suggestions? Jul 09 19:20:16 seems like that should work... Jul 09 19:32:43 i can't find any examples tho, so you're probably right... Jul 09 19:44:43 hi. I'm trying to get a kernel to boot, but I cant make u-boot load it. u-boot spits out both "load address" and "entry point" to 00008000, but I set UBOOT_ENTRYPOINT/LOADADDRESS = "0x80008000" Jul 09 19:44:49 in my machine config Jul 09 19:44:52 any idea what is wrong? Jul 09 19:45:43 I'm using v2013.01 u-boot and u-boot-mkimage from oe-core Jul 09 19:46:13 as well as a linux-yocto-3.4 kernel, with my own config Jul 09 22:48:35 this is rare, but i'm actually sorry i tried to help that guy... Jul 09 22:49:29 i think he's more stubborn that all 3 of my kids put together Jul 09 22:49:36 HAHAHAH Jul 09 22:50:04 hah Jul 09 22:50:28 :P Jul 09 22:50:32 I feel much less stress after /ignore Jul 09 22:51:10 i'm sure that was one of the very first irc commands Jul 09 23:00:45 i mean, i'm more than happy to rag on kergoth just because, but that was starting to grate on me... Jul 09 23:01:48 * kergoth chuckles Jul 09 23:02:55 kergoth he's now planning on redoing the bitbake parser.. since it's obviously broken Jul 09 23:03:32 hah. the file format and parser have their issues, but this is not one of them Jul 09 23:03:58 I re-pointed out the include/require logic and he said that's broken cause it's not a simple design.. :) Jul 09 23:04:24 boggles the mind Jul 09 23:04:38 I agree a ~/.yocto-rc is necessary, I will now argue about it here, cause the other channel is busy Jul 09 23:04:50 ^_^ Jul 09 23:04:58 HAHAHAHA Jul 09 23:05:02 ..unless cookies Jul 09 23:05:38 c00kiemon5ter: hang around my house for a while and whine about chocolate chip cookies... Jul 09 23:05:52 x) Jul 09 23:07:14 ....if this fetchall ever finishes, I can get back to work and stop playing with the troll Jul 09 23:07:48 fray: or keep him busy until the sun comes up Jul 09 23:08:11 at which point we'd need to fit you with a straight jacket Jul 09 23:09:22 * ant_home got goosebumps #beagle trolling feelings Jul 09 23:10:59 * kergoth thinks back to how little patience he had for newbs when the project was first starting, amazing it ever took off ;) Jul 09 23:11:57 i don't remember you being that much different Jul 09 23:12:12 maybe we've both mellowed some... Jul 09 23:12:41 * Daemon404 isnt sure he'd call bitbake's design good... but not 'broken' Jul 09 23:12:42 * Daemon404 runs Jul 09 23:12:43 now it takes someone like this guy to piss me off, rather than someone just asking a question without reading the damn manual :) Jul 09 23:15:18 I wonder if I can convince him w/o convincing casual observers that this stuff is all crap so he goes away Jul 09 23:15:22 kergoth: HEY, bitBroke IS BORKED.WILL YOU FIX IT FOR ME PLZZ Jul 10 00:07:54 this is still going; next bug report will be an RFC for IRC channel communication guidelines Jul 10 00:08:19 I litterly laughed out loud on that Jul 10 00:08:32 Jul 05 12:05:56 but I do not think it is Yocto related now. Jul 10 00:08:35 for a troll he's not very fully.. what that was a zinger.. :0 Jul 10 00:08:41 Jul 05 12:06:06 I am getting rid of Yocto for the time being. Jul 10 00:08:48 'fully' -- funny Jul 10 00:08:49 jeez..almost Jul 10 00:08:56 please.. please... Jul 10 00:09:35 he'll report to Jeff; Jeff will take care of everything; fray you'll be punished Jul 10 00:09:46 you'll be put in charge of feeding the trolls Jul 10 00:10:03 I'll send back the yocto t-shirt from Barcelona ;) Jul 10 00:10:06 what did i miss? Jul 10 00:10:29 Jul 10 00:10:39 he's going to report all of us to Jefro for being mean Jul 10 00:11:32 it's a one-man show, scary... Jul 10 00:11:42 btw, I am new here, and the #yocto channel is way friendly. He is probably new to IRC, cause there are too many channels out there - at least half of which I'm joined - that would have banned him long ago Jul 10 00:11:51 first one that comes to mind is ##c Jul 10 00:12:03 he claims he's not new to IRC.. Jul 10 00:12:14 yeah, I saw that, that's why I made that remark ^ Jul 10 00:12:25 so I can only guess he's ignorant and won't listen, a troll.. or attempting to flamebait the channel Jul 10 00:12:26 please don't use ##c as an example... Jul 10 00:12:36 lol Jul 10 00:12:48 that channel is as wonky as the language... Jul 10 00:12:56 he would make a record join-to-ban in ##c Jul 10 00:13:00 would that get you banned? ;) Jul 10 00:14:03 I still suspect trolling because he gets "defensive".. says he's going to take his ball and go home.. then less then a minute later he starts rambling again.. Jul 10 00:14:05 ##c is scrict for sure; you won't have fun if you don't speak the standard Jul 10 00:14:05 we beat #beagle 10-1 wrt trolling in 2013 Jul 10 00:14:13 alright, i now have test prints in 12 different languages Jul 10 00:18:20 bluelightning, crazy man, inviting for tomorrow... Jul 10 00:18:57 time to sleep, anyway Jul 10 00:19:00 gn Jul 10 00:19:32 someone should respond, that he should not give up and he should have good faith when someone calls him a troll Jul 10 00:22:49 if he doesn't get that part already, then i'm not sure telling him will do any good... Jul 10 00:47:06 * mr_science goes home... **** ENDING LOGGING AT Wed Jul 10 02:59:59 2013