**** BEGIN LOGGING AT Sat Jun 18 02:59:56 2011 Jun 18 05:45:15 Hi guys, what's the current state of bitbake interactive and is any work being done on supporting or any alternatives? Jun 18 10:14:58 hi Jun 18 10:15:04 I'm blocked on gcc-runtime Jun 18 10:15:12 it always fails with: Jun 18 10:15:24 cannot find -lm Jun 18 10:15:28 on oe-core Jun 18 10:15:32 what can I do Jun 18 10:15:38 I cleaned the toolchain Jun 18 12:13:15 Running into issues fetching source when compiling fsogsmd. Any ideas? http://pastie.org/2086915 Jun 18 12:15:23 Lopi: not sure but it looks to me as if SRCREV is undefined for the fsogsmd recipe in your configuration Jun 18 12:19:09 bluelightning: SRCREV = "${FSO_CORNUCOPIA_SRCREV}" Jun 18 12:20:57 actually the items it's complaining about are libfso-glib and fso-specs Jun 18 12:21:06 both of those recipes define their own SRCREV Jun 18 12:21:16 so I have to say I'm not sure what could be wrong Jun 18 12:21:31 maybe try bitbake -e libfso-glib | grep SRCREV Jun 18 12:21:39 see what it reports Jun 18 12:23:54 too much data to sift through Jun 18 12:24:02 why wouldnt I look at the .bb? Jun 18 12:24:36 Lopi: other things can have an effect on the variables, such as distro config or even stuff in local.conf if you've added it Jun 18 12:24:50 I see Jun 18 12:24:53 bitbake -e shows the whole environment that is actually what is being used at build time Jun 18 12:25:28 well, the device I'm building for doesn't have any special srvrev Jun 18 12:25:52 Lopi: you aren't by any chance doing include or require on fso-autorev.inc anywhere? (nothing does by default) Jun 18 12:26:10 bluelightning: yeah I'm doing require fso-autorev.inc in local.conf Jun 18 12:26:22 right, so that's probably the problem Jun 18 12:26:40 I don't know much about AUTOREV but it's at least not functioning for you :/ Jun 18 12:27:39 been trouble shooting compile errors for the past 90 minutes instead of testing my new plugin for the baseband -_- Jun 18 12:28:24 can I do source ~/path/to/devshell without autorev? Jun 18 12:29:22 hmm... JaMa is not around, I'm sure he could advise on these issues Jun 18 12:29:51 ah, so you don't know how to test new fsogsmd plugins? Jun 18 12:30:12 no, I've never built FSO myself... I know some of the people involved, but that's it Jun 18 12:30:41 GNUtoo told me to require autorev then source dev shell and compile the it from there Jun 18 12:30:42 perhaps if AUTOREV is not working you could just set it to some fixed revision Jun 18 12:32:14 I'm just going to take a break, this is silly ;p Jun 18 14:22:03 otavio: hi, fyi I've established the nativesdk problem is not gcc 4.6.0 related Jun 18 15:29:27 bluelightning: and did you figure why it has started to happen recently? Jun 18 16:17:55 otavio: no, I have got no further Jun 18 16:41:02 bluelightning: right Jun 18 16:41:26 bluelightning: I am testing a full rebuild of our system and will then focus on qt4 Jun 18 16:41:40 bluelightning: the bbappend stuff you had, is ready? Jun 18 16:41:46 otavio: ok great Jun 18 16:42:01 otavio: almost :) Jun 18 16:42:17 bluelightning: I am looking forward to it Jun 18 16:42:32 bluelightning: btw, nativesdk of qt4 4.7.2 fails to compile here too Jun 18 16:42:51 bluelightning: so it seems you don't need to wait for fix 4.7.3 to send it Jun 18 16:42:59 bluelightning: it is not a regression after all Jun 18 16:43:08 otavio: right Jun 18 16:43:56 bluelightning: I might be able to work with it since I depend on it too Jun 18 16:44:06 bluelightning: help with it... sorry Jun 18 16:44:20 that would be much appreciated :) Jun 18 17:32:15 bluelightning, Hope this isn't a silly question but when I look at the compiler logs (log.do_compile) for a recipe, I don't see the same commands (such as make all..) in the do_compile function of the recipe source, I just see the output of those commands. Is there a way for me to do this? Jun 18 17:32:44 joelagnel: perhaps run.do_compile? Jun 18 17:34:21 actually no that's not going to contain what you want Jun 18 17:35:34 I can only say log.do_compile contains the commands and output in the logs for my builds... Jun 18 17:36:09 run.do_compile are the commands, and log.do_compile are the results of running those commands Jun 18 17:36:30 not all log.do_compile will containt he command itself.. it depends on if the command is shell or python.. and if its shell -- if set +x is set Jun 18 17:37:42 bluelightning, fray oh I see, ok thanks. run.do_compile looks like a shell script. I'm guessing this is the final script that is executed with all the variables substituted? Jun 18 17:37:53 joelagnel: yes Jun 18 17:38:07 yes, thats the intention.. Jun 18 17:38:19 nice Jun 18 17:38:32 a few odd-ball recipes run multiple scripts with the same name (I don't remember why).. if it does.. look at the version witht he PID.. since the pids change between runs Jun 18 17:38:53 the run.do_compile and log.do_compile is a symlink to the last command run Jun 18 17:39:21 If I were to just execute this script, without starting bitbake, would it work the same way as it would when bitbake ran it? Jun 18 17:39:28 the pids and links is also a way to show multiple runs Jun 18 17:39:54 joelagnel if you are in the bitbake devshell, yes it should.. Jun 18 17:40:43 (I don't use the bitbake devshell normally, so I have to always look up how to get into it) Jun 18 17:40:57 fray, I think the option is "-c devshell" Jun 18 17:41:17 likely thats it Jun 18 17:42:07 fray, but I get a cannot open display error Jun 18 17:42:27 you need to setup the TERMCMD and related settings in your local.conf Jun 18 17:42:54 fray, oh ok my bad, will check it out, thanks Jun 18 17:42:56 go to the end of the local.conf and you should see some examples.. uncommend and change to the right version Jun 18 17:43:03 (I usually use screen myself Jun 18 17:44:55 fray, did you mean bitbake.conf? Jun 18 17:45:04 my local.conf doesn't have TERMCMD or examples Jun 18 17:46:42 sorry.. I thought I was in a different channel.. Jun 18 17:46:49 Umm the example I have is: Jun 18 17:47:05 # If you do not use (or have installed) gnome-terminal you will need to Jun 18 17:47:05 # uncomment these variables and set them to the terminal you wish to use Jun 18 17:47:05 # when resolving patches which cannot be applied Jun 18 17:47:05 # Supported shell prefixes for *_TERMCMD and *_TERMCMDRUN ARE: Jun 18 17:47:05 # GNOME, SCREEN, XTERM and KONSOLE Jun 18 17:47:06 #TERMCMD = "${KONSOLE_TERMCMD}" Jun 18 17:47:06 #TERMCMDRUN = "${KONSOLE_TERMCMDRUN}" Jun 18 17:47:14 that should work with oe-dev as well Jun 18 17:51:08 fray, thanks, I saw the same example in bitbake.conf. Trying screen now Jun 18 17:55:22 fray, with TERMCMD as screen, it hangs for a long period of time at "task do_devshell: Started" . I think saw this the other day as well Jun 18 17:55:57 it's not hung.. the screen has started.. open another terminal (x, ssh, etc) and use screen to go find it.. Jun 18 17:56:10 (it's sitting waiting for you to use the devshell and close it) Jun 18 17:56:52 sigh :)) Jun 18 17:57:24 * fray notes this is a complaint he has with the devshell... it's not obvious it's waiting on -you-.. it does look hung Jun 18 17:58:46 fray, wow a whole bunch of env vars set. Guessing this is the toolchain and stuff, very neat :) Jun 18 18:01:58 fray, do you know if there are any patches I can apply to get interactive-mode working? Jun 18 18:08:16 fray, sorry I will like to take that back and ask it in a more polite way :) Jun 18 18:08:31 does bitbake -i work for you? Jun 18 18:08:58 I've never used it.. Jun 18 18:09:22 I get no such option -i Jun 18 18:11:23 fray, yeah, I think it used to be there in older versions of bitbake though Jun 18 18:11:40 joelagnel: set PATCHRESOLVE = "user" in your local.conf Jun 18 18:11:54 in oe-dev it defaults to noop Jun 18 18:12:21 * fray notes he sets PATCHRESOLVE = "noop" cause I can't stand the devshell when a patch doesn't apply.. ;) Jun 18 18:12:36 * bluelightning does the same :) Jun 18 18:14:11 lightning here is getting pretty intense.. hopefully I won't lose power Jun 18 18:15:09 ugh -- in oe-core the qt4 package has a custom debug package rule set.. which doesn't work like the rest of oe-core.. Jun 18 18:15:19 * fray isn't going to spend the time to fix it.. at least not today Jun 18 18:16:15 fray: there's definitely more cleanups that could be done on the qt4 recipes Jun 18 18:16:47 ya.. I see why it has a customs et.. they're trying to breakup qt4 into logical components.. but the debug stuff (default) works on a package as a whole.. Jun 18 18:17:07 so either we need to fix the "breakup" code.. or change the behavior of the qt4 to have one -large- debuginfo package Jun 18 18:17:13 (later is certainly easier) Jun 18 18:18:39 bluelightning, how would PATCHRESOLVE help in my situation? Jun 18 18:19:42 does it have something to do with do_patch or... Jun 18 18:20:51 if do_patch fails, patchresolve will stop (in the devshell) and let you fix it Jun 18 18:21:36 oh I see, very cool.. Jun 18 18:23:30 * joelagnel wonders why bitbake -i is not supported any more Jun 18 18:23:36 I'm using a workflow like this: http://www.slimlogic.co.uk/2011/05/openembeddedangstrom-kernel-workflow/ Jun 18 18:23:53 it takes a long time each time the parsing has to happen Jun 18 18:24:45 long In the sense a few seconds :), it would be nice if the parsing/loading-of-cache could be done just once for a workflow Jun 18 18:28:31 sorry, I was trying to say that the "parsing" between different commands in the workflow seems repetitive Jun 18 18:30:26 I'm compiling fsogsmd to test my plugin for the iPhone3G's baseband, but I'm running into some compile issues. Any ideas? http://pastie.org/private/etglrjh9cvuq2193qxwp6w Jun 18 18:39:49 otavio: I've just applied the patch from the qt bugtracker, it works so I'm submitting it Jun 18 18:40:00 obi Jun 18 18:40:07 bluelightning: it is awesome :-) Jun 18 18:40:49 bluelightning: please push it to contrib ... so i can merge it before Jun 18 18:41:32 otavio: you want to test it first? Jun 18 18:41:56 bluelightning: it works too Jun 18 18:44:35 otavio: contrib paule/qt4-nativesdk-fix Jun 18 18:45:03 that's oecore contrib Jun 18 18:45:26 although there's a poky-contrib paule/qt4-nativesdk-fix1 which is the same thing Jun 18 18:45:30 on top of poky Jun 18 18:45:30 bluelightning: did intltool failed for you? Jun 18 18:45:32 Lopi, hi Jun 18 18:45:42 otavio: not that I noticed... Jun 18 18:46:13 GNUtoo: hey Jun 18 18:47:56 Lopi, did you solve your fso-specs issue? Jun 18 18:48:33 GNUtoo: only if I remove autorev from local.conf Jun 18 18:48:47 GNUtoo: then orc seg faults when it tries to make Jun 18 18:49:55 ah? Jun 18 18:50:10 remove autorev from fso-specs only then Jun 18 18:50:42 from the bb? Jun 18 18:51:00 from the autorev inc Jun 18 18:52:11 fray, hi, I've cannot find -lm on gcc-runtime on oe-core Jun 18 19:00:25 otavio: have sent all patches, incl meta-oe bbappends Jun 18 19:02:18 bluelightning: meta-oe bbappend, please push it to meta-openembedded-contrib so I can pull them Jun 18 19:02:36 otavio: already done, see branch paule/qt4-bbappends Jun 18 19:03:04 otavio: sorry paule/qt4-bbappend , no s Jun 18 19:13:59 note that I tried to clean the toolchain in oe-core Jun 18 19:14:07 should I rebuild oe-core from scratch? Jun 18 19:16:56 is there anyone that know a bit oe-core? Jun 18 19:17:15 i know oe-core pretty well, but nothing about the toolchain stuff. what exactly are you trying to do? Jun 18 19:18:51 compile an image Jun 18 19:18:53 shr-image Jun 18 19:19:00 but I'm stuck here: Jun 18 19:19:19 when it compiles gcc-runtime Jun 18 19:19:25 it cannot link against libm Jun 18 19:21:02 it seems(not sure) that gcc-runtime is compiled before the eglibc Jun 18 19:21:35 ah now I've | configure: error: C preprocessor "arm-oe-linux-gnueabi-gcc -E --sysroot=.../sysroots/om-gta02" fails sanity check Jun 18 19:21:55 so my question is now: Jun 18 19:22:02 how do I clean the toolchain? Jun 18 19:22:09 so I've a clean restart Jun 18 19:22:19 my clean toolchain does that: Jun 18 19:22:54 bitbake -c clean gcc gcc-cross gcc-cross-intermediate gcc-runtime binutils binutils-cross eglibc eglibc-initial Jun 18 19:23:37 why not just rm -rf the build dir? Jun 18 19:23:48 with sstate it'll just use the cached bits where appropriate Jun 18 19:23:51 from sstate-cache Jun 18 19:23:59 ok Jun 18 19:23:59 wiping tmpdir isn't particularly painful Jun 18 19:25:35 ok I'll do that right now Jun 18 19:33:28 GNUtoo: what do I do if a compile is segfaulting? Jun 18 19:33:47 pastebin the segfault Jun 18 19:34:06 k Jun 18 19:34:27 http://pastie.org/private/egc1xwi2vka5dnmog6gq Jun 18 19:37:46 I've no idea beside recompiling orccc with ggdb3 Jun 18 19:37:55 look in bitbake.conf for that Jun 18 19:38:32 inside oe dir? Jun 18 19:39:38 bitbake.conf is in the metadata Jun 18 19:39:55 /home/lopi/iX/ix-build/bitbake/conf/bitbake.conf Jun 18 19:39:55 /home/lopi/iX/ix-build/openembedded/conf/bitbake.conf Jun 18 19:40:07 /home/lopi/iX/ix-build/openembedded/conf/bitbake.conf Jun 18 19:40:19 what do I edit? Jun 18 19:40:24 I'm looking through it Jun 18 19:41:17 not sure Jun 18 19:41:20 maybe BUILD_OPTIMIZATION Jun 18 19:41:39 yes should be that Jun 18 19:41:42 else since it's -g Jun 18 19:41:45 already Jun 18 19:41:51 you could try gdb on it Jun 18 19:42:04 I've a big problem with emacs-x11 tough, gdb segfault Jun 18 19:42:22 s/gdb/qemu/ Jun 18 19:42:26 this is dumb, I don't understand why I can't just compile the plugin separately Jun 18 19:42:34 what plugin? Jun 18 19:42:40 you need dependencies Jun 18 19:42:43 I modified singleline Jun 18 19:42:52 in fsogsmd Jun 18 19:42:55 yes but fsogsmd has dependencies Jun 18 19:43:03 you have 2 options Jun 18 19:43:07 I already have it on my device tho Jun 18 19:43:10 *compile fsogsmd inside oe Jun 18 19:43:15 *compile it outside oe Jun 18 19:43:26 but in any case you need to compile fsogsmd Jun 18 19:43:42 if it's in oe you could do the modifications inside oe Jun 18 19:43:55 if you do it outside you've got some options: Jun 18 19:43:58 *devshell Jun 18 19:44:13 *use the temp compile script of the one compiled in oe Jun 18 19:45:45 I don't understand why sources that segfault are pushed upstream Jun 18 19:45:52 <-- frustrated Jun 18 19:45:56 been at this for several hours Jun 18 19:46:17 try to find a way to compile fsogsmd Jun 18 19:46:21 by whatever means Jun 18 19:46:31 if you already have one compiled Jun 18 19:46:36 it seems intltool is broken Jun 18 19:46:39 you could use that one Jun 18 19:46:40 or something broke it Jun 18 19:46:51 I don't even know what orcc is Jun 18 19:46:57 maybe mickey|daddy would know? Jun 18 19:47:03 I code for 30 mins then fight with oe for hours >< Jun 18 19:47:34 that's because no one was there for helping you Jun 18 19:47:52 you even have a third option Jun 18 19:47:53 yeah :/ Jun 18 19:48:00 run fsogsmd on your laptop/desktop Jun 18 19:48:06 and forward the serial Jun 18 19:48:14 so you don't have to deal with oe Jun 18 19:48:49 since you're currently frustrated that may be the best way....but Jun 18 19:49:09 watch out for dependencies Jun 18 19:49:16 for instance if it require a recent vala Jun 18 19:49:44 orcc is the oil runtime compiler, a tool for writing SIMD programs (like SSE) in a generic manner Jun 18 19:50:03 ah ok, but what is oil? Jun 18 19:50:06 so is orcc an fsogsmd dependency? Jun 18 19:50:17 Lopi, it's a dep of what you bitbaken Jun 18 19:50:35 Lopi, else you could try to use your distro orcc Jun 18 19:50:36 OIL stands for Optimized Inner Loops Jun 18 19:50:39 ok Jun 18 19:51:09 bleh I might just give up for the day :/ Jun 18 19:51:17 what about trying to compile fsogsmd on your laptop and forward the serial from the iphone to the laptop Jun 18 19:51:31 going on too long while frustrated doesn't work Jun 18 19:51:34 using networking Jun 18 19:51:54 you'll just create a bunch of horrible hacks Jun 18 19:51:55 well I'm pissed because I was really excited about testing my plugin Jun 18 19:52:03 been working on it for a long time.. Jun 18 19:52:09 Lopi, maybe I could compile it for you Jun 18 19:52:40 it's 2:51 PM and I finished my changes to the plugin at around 8:00 a.m. Jun 18 19:53:00 I'm fine with that if you want to do it Jun 18 19:53:43 push the sources somewhere Jun 18 19:53:46 like on gitorious Jun 18 19:53:57 it's one plugin Jun 18 19:54:23 pastebin then Jun 18 19:55:43 http://pastie.org/private/enk8xsqplu5aryq0idaq Jun 18 19:56:26 I don't even know if it's going to compile properly. I wasn't very sure of a certain part Jun 18 19:56:38 I'll compile it and see Jun 18 19:57:08 if it doesn't I'll ask mickey or morphis Jun 18 19:58:04 I'm worried about lines 122-123 Jun 18 19:58:45 bluelightning: I am rebuilding oe-core image from scratch and will test your changes at same time Jun 18 19:58:52 bluelightning: let's see how it goes Jun 18 20:01:18 there is even a forth option: use the shr-makefile with its chroot Jun 18 20:02:18 free(cmd);, Jun 18 20:02:20 remove the , Jun 18 20:03:42 oh Jun 18 20:03:45 I didn't read the code Jun 18 20:04:16 first add a ) Jun 18 20:04:53 let me try to fix it Jun 18 20:06:31 okay Jun 18 20:06:36 let me know the changes when you're doing Jun 18 20:06:44 done* Jun 18 20:44:25 hmmm Jun 18 20:45:33 anyone have an external toolchain recipe that works with crosstool-ng, or will our generic one work? Jun 18 20:45:36 * kergoth checks the recipes Jun 18 21:32:39 GNUtoo: heh it segfaults Jun 18 21:34:36 kergoth: btw, about building from git directly Jun 18 21:35:02 ALT Linux distro doing this about 2 years Jun 18 21:35:23 infrastructure is accessible Jun 18 21:35:30 gpl'ed iirc Jun 18 21:35:45 they are building rpm's from git Jun 18 22:13:34 Jay7: many, many distros are doing it now :) Jun 18 22:13:37 git.debian.org, etc Jun 18 22:13:40 just not fully Jun 18 22:16:39 sounds cool though, i should check it out Jun 18 22:16:47 seen project-builder, for upstreams? Jun 18 22:16:49 looks quite interesting Jun 18 22:36:16 btw I've that now: Jun 18 22:36:22 checking for suffix of object files... configure: error: in `/home/gnutoo/port4/embedded/oe-core/oetmps/shr/work/armv4t-oe-linux-gnueabi/eglibc-2.13-r2+svnr14157/build-arm-oe-linux-gnueabi': Jun 18 22:36:33 configure: error: cannot compute suffix of object files: cannot compile Jun 18 22:36:37 I'll look at config.log Jun 18 22:39:14 armv4t-oe-linux-gnueabi.gcc-cross-intermediate/gcc/arm-oe-linux-gnueabi/4.5.3/cc1: error while loading shared libraries: libmpc.so.2: cannot open shared object file: No such file or directory Jun 18 22:39:16 hmmm Jun 18 22:39:27 what is that lib? Jun 18 22:41:06 * kergoth has no idea Jun 18 22:43:02 ah ok Jun 18 22:43:16 MPC is a C library for the arithmetic of complex numbers with Jun 18 22:43:16 arbitrarily high precision and correct rounding of the result. It is built upon and follows the same principles as Mpfr. Jun 18 22:43:33 I hope it's not part of gcc **** ENDING LOGGING AT Sun Jun 19 02:59:57 2011