**** BEGIN LOGGING AT Fri Nov 04 02:59:56 2005 Nov 04 03:01:01 RP: a (probably stupid) question - with kernel 2.6, are wmmx extensions enabled? Nov 04 03:01:26 I keep hearing people talk about them "not suspending properly". I've yet to see it though (unless people mean when a process fails to freeze? Nov 04 03:01:37 Bernardo: No, they're not and we probably should enable that Nov 04 03:01:46 RP: this is purely user error, not code Nov 04 03:01:59 XorA: ok, fair enough :) Nov 04 03:02:17 what do the iwmmx extensions do? Nov 04 03:02:31 afaik our userspace can't handle it due to softpa being used Nov 04 03:02:45 RP: I left turned on and closed :-d Nov 04 03:02:56 koen: mmx/sse for arm cpus Nov 04 03:03:14 they help when you want to watch videos on the akita... :) Nov 04 03:03:34 RP: they're off because of softfpu? Nov 04 03:03:42 03rpurdie 07org.oe.dev * r208a9e3e... 10/: Nov 04 03:03:43 linux-oz-2.6: dev branch updates. Update hx2750 patches to match recent mainline Nov 04 03:03:43 changes. Alter patch order to match planned mainline submission. Make final Nov 04 03:03:43 cleaup of battery/power patches ready to push to arm patch system (started). Nov 04 03:03:47 03frederic 07org.oe.dev * refc301dc... 10/: petitepainture: dialogs now work (added patch) in 1.5 Nov 04 03:03:58 RP: I have no soc-pxa2xx-soc-corgi Nov 04 03:04:17 RP: actually ignore, I cannot read this morning either Nov 04 03:04:27 RP: that'll be the module I missed then Nov 04 03:09:40 brb Nov 04 03:14:02 Bernardo|Away: I'm not sure if they conflict with softfpu or not... Nov 04 03:15:57 koen: I'm not sure softfpu conflicts with iwmmx... Nov 04 03:16:31 If fact I'd say not Nov 04 03:16:46 softpa Nov 04 03:16:52 not softfpu Nov 04 03:17:39 RP: http://pastebin.ca/27583 Nov 04 03:18:13 RP: iirc softpu is actually required for iwmmx to work Nov 04 03:18:48 koen: softpa being what? Nov 04 03:19:12 XorA: oops. I suspect your module load oirder caused that... Nov 04 03:19:22 The load order admittedly needs work Nov 04 03:19:24 softfpa, the softfloat variant we use on OZ Nov 04 03:20:40 koen: ah, softfpa... Nov 04 03:21:09 heh, I just noticed the typo Nov 04 03:21:24 but enabling it in the kernel shouldn't cause problems Nov 04 03:21:43 our userspace just can't do usefull things with it Nov 04 03:22:33 yes. I think I'll enable it in the kernel Nov 04 03:23:09 Angstrom should be able to handle it with eabi (which uses (soft)vfp) Nov 04 03:23:29 and staticly built software should work too Nov 04 03:23:34 koen: mplayer has iwmmx assembler in it, so enabling the extensions would be good Nov 04 03:26:27 koen: I can't find evidence of a problem when combining softfpa and iwmmxt ... Nov 04 03:26:47 RP: even better Nov 04 03:27:02 I remember the fpa format being troubles, but it could be limited to hardfpa Nov 04 03:27:48 hardfpa is much more likely to cause problems than softfpa Nov 04 03:30:19 koen: how was your try to get eabi toolchain? Nov 04 03:30:38 glibc didn't compile Nov 04 03:30:43 but bero is working on it Nov 04 03:31:08 http://handhelds.org/hypermail/oe/42/4241.html Nov 04 03:32:25 thx Nov 04 03:32:34 I had to fetch mail Nov 04 03:34:05 'morning all ( according to ugt ) :) Nov 04 03:34:20 'lo Nov 04 03:35:08 mattina Pigi Nov 04 03:35:29 hrw, heh. mattina to you Nov 04 03:36:36 I have a question for the oe guru's. I'm trying to slim down a bit an image for my own purpose. Nov 04 03:36:56 Basically the most of removing is done on the h3900.conf ( local ) Nov 04 03:37:37 hi Pigi Nov 04 03:37:44 mornin' Nov 04 03:37:45 what's happening is that if I modify the h3900.conf ( local ), the changes don't take effect until I do remove the whole tmp/ stuffs Nov 04 03:38:10 hi RP. I haven't sleep too well tonight thinking at your bug. Nov 04 03:39:11 Is the behavior I've described correct or I'm missing something ? Nov 04 03:40:02 Pigi: Sorry about that :-/ Nov 04 03:40:50 RP: I would like to discuss this problem a bit further. Would you be there in the afternoon ? Nov 04 03:41:09 Pigi: I should be, yes Nov 04 03:41:17 good. Nov 04 03:43:08 Pigi: #define 'modify the h3900.conf' Nov 04 03:43:23 Pigi: added/removed packages/modules? Nov 04 03:43:59 hrw|work, yes, exactly. Basicaly I'm removing every ( at least for me ) unneeded modules and packages as openswan Nov 04 03:44:03 ipv6 Nov 04 03:44:07 and so on Nov 04 03:44:11 Pigi: rebuild task-bootstrap Nov 04 03:44:42 Pigi: task-bootstrap depend on everything which is in machine/distro config. all images depend on task-bootstrap Nov 04 03:44:48 hrw|work, that's a nice advice. Nov 04 04:00:22 koen: Any progress on leaning on the gpe people about the font sizes? Nov 04 04:00:40 yes Nov 04 04:00:53 I've convinced them it's a problem Nov 04 04:01:10 koen: That's a good first step I guess :) Nov 04 04:01:19 I think I'll add a temp hack in OE to overwrite the default Xsettings for large DPI machines Nov 04 04:01:38 koen: That would be appreciated Nov 04 04:01:51 * RP would like to build a working gpe image Nov 04 04:02:10 the notion 'big screen != big dpi' had to sink in a bit Nov 04 04:02:44 :) Nov 04 04:05:30 A decent fstab makes spitz so much more sane... Nov 04 04:08:02 RP: do you have a modules order for c7x0 that works? Nov 04 04:09:16 XorA: http://www.rpsys.net/openzaurus/temp/inssound Nov 04 04:09:24 change 8750 for 8731 Nov 04 04:10:53 hmm.. I dont have corgi one too Nov 04 04:13:15 but kernel building on the way Nov 04 04:13:56 btw.. from where does OE take kernel modules descriptions? Nov 04 04:16:43 from modinfo? Nov 04 04:17:01 if not, using modinfo would be a good starting point Nov 04 04:17:38 yep Nov 04 04:18:22 some modules on my Z have names from modinfo, some have just 'MODNAME kernel module', some does not have any (but have description in modinfo) Nov 04 04:18:51 kernel.bbclass should handle it Nov 04 04:19:19 cmd = "PATH=\"%s\" %sobjcopy -j .modinfo -O binary %s %s" % (bb.data.getVar("PATH", d, 1), bb.data.getVar("HOST_PREFIX", d, 1) or "", file, tmpfile) Nov 04 04:19:33 complete gibberish to me Nov 04 04:24:04 03rpurdie 07org.oe.dev * r4de71cbd... 10/: linux-oz-2.6-git3: Enable iwmmxt for pxa27x devices Nov 04 04:24:08 03rpurdie 07org.oe.dev * r1bfddc7c... 10/: linux-oz-2.6-git3: Update defconfigs to ensure pcmcia cf wifi modules get built. 2.6.14-rc1 could really use a similar update. Nov 04 04:24:12 03rpurdie 07org.oe.dev * r69065044... 10/: base-files: Correct fstab for spitz. Its slightly 2.6 specific now but not enough to break 2.4. Nov 04 04:56:16 RP: cheers, that script does the business Nov 04 04:57:25 RP: wow, booting with rotated screen works in my akita now Nov 04 04:57:44 evrytime I compile a new image I found a lot of things improved, GREAT JOB Nov 04 05:05:28 RP: do you need to mess with anything else to unmute? Nov 04 05:12:38 RP: is suspend working in the akita with 2.6? Nov 04 05:15:34 It works with Spitz. Nov 04 05:21:42 katossi: yes, it should do Nov 04 05:22:05 XorA: You might need to mess with the mixer and/or the sysfs files :-/ Nov 04 05:22:20 That really narrows it down, I know ;-) Nov 04 05:31:50 XorA: install alsamixer and alsaconf Nov 04 05:42:07 <_law_> koen: hey, saw your gaim screen any ideas what i should do? Nov 04 05:42:17 <_law_> koen: did sound work for you? Nov 04 05:42:46 like the bugzilla entrie says, remove all patches from DL_DIR Nov 04 05:44:18 hrw|work: that is what I am using, it produces no sound Nov 04 05:53:38 how can i write in a .bb file that an envirnment variable have to beet set for the build Nov 04 06:16:01 re Nov 04 06:18:46 what does base.bbclass descripe? Nov 04 06:27:00 I have noticed that gpe-bluetooth doesn't RDEPENDS from libmimedir, (neither DEPENDS though) which should be really needed if you don't install gpe-syncd. Nov 04 06:27:09 Could this be seen as a bug ? Nov 04 06:28:14 Pigi: not the RDEPENDS (as this is gathered automatically as well) Nov 04 06:28:39 Pigi: libminedir is a build time dep, it should be either directly or indirectly in the DEPENDS Nov 04 06:29:28 well, libmimedir is also needed by gpe-bluetooth to work Nov 04 06:29:29 root@h3900:/usr/lib/ipkg/info# /usr/bin/gpe-bluetooth Nov 04 06:29:29 /usr/bin/gpe-bluetooth: error while loading shared libraries: libmimedir-0.2.so.0: cannot open shared object file: No such file or directory Nov 04 06:30:05 but on the image I'm working on, there is no other packages that RDEPENDS on it, so it doesn't get installed by the image itself. Nov 04 06:30:54 and is DEPENDed by gpe-beam, which goes installed but don't install libmimedir ( as it does not [R]depends on it) Nov 04 06:31:50 Pigi: well as I tried to say, if it links against libmimedir-0.2.so.0 it will automatically land in RDEPENDS Nov 04 06:32:40 so there should be a bug somewhere, as it links against libmimedir but doesn't install it. ( I haven't yet understood how oe works :) ) Nov 04 06:32:58 Pigi: neither did we Nov 04 06:33:06 heh Nov 04 06:33:08 Pigi: DEPENDS is for build time dependencies Nov 04 06:33:30 and that's ok, and RDEPNEDS is what goes in Depends: for the control file Nov 04 06:33:37 Pigi: and RDEPENDS for runtime stuff Nov 04 06:33:39 Pigi: right Nov 04 06:33:55 then, ( just to explain me better ) Nov 04 06:33:57 Pigi: for linked shared libraries it tries to find out which package provides this library Nov 04 06:34:16 and then adds this package name to the RDEPENDS automatically Nov 04 06:34:39 if gpe-bluetooth doesn't DEPEND on libmimedir, I should at least add it in DEPENDS . Nov 04 06:35:00 Pigi: why? Nov 04 06:35:17 Pigi: if gpe-bluetooth does not need libmimedir to build (the package) Nov 04 06:35:26 it should not have libminedir in DEPENDS Nov 04 06:35:28 as probably, it can use it at link time when doing a build because gpe-beam DEPENDS on it Nov 04 06:35:58 well I do not know the Dependency relation of all GPE packages *sorry* Nov 04 06:36:05 zecke, I feel dumb: when running by hand gpe-bluetooth, it complains about the missing lib Nov 04 06:36:36 but it is not in DEPENDS for gpe-bloodtooth Nov 04 06:37:05 I should ask the maintainer.... Nov 04 06:37:13 Pigi: well did you create the package yourself? Nov 04 06:37:22 No, pb_ did it. Nov 04 06:37:38 I'm just tring to create a minimal gpe image for my needs. Nov 04 06:37:44 oris_wolfbane: hi Nov 04 06:37:59 using the framework we actually have. Nov 04 06:38:03 zecke, hey Nov 04 06:38:21 ( I would like to remove all the ipv6 and openswan and ipsec and ppp stuff ) Nov 04 06:47:31 How can I set an environment variable for the compilation? Nov 04 06:52:00 export FOO? Nov 04 06:52:06 hey guys, when ever i try and build somthing i get "Fatal error, aborting. Nov 04 06:52:07 anoncvs:anoncvs: no such user Nov 04 06:52:07 cvs checkout: authorization failed: server anoncvs.handhelds.org rejected access to /cvs for user anoncvs:anoncvs Nov 04 06:52:07 " Nov 04 06:52:16 any ideas? thanks in advance? Nov 04 06:52:46 okay and when I write that in the do_configure() funktion is it visible in the do_compile()? Nov 04 06:55:25 oris_wolfbane, could it be you have a ~/.cvspass with some wrong information ? Nov 04 06:55:36 * oris_wolfbane goes and looks Nov 04 06:59:40 ive just backed the file up and then wiped it.. seeing if it makes a difference Nov 04 07:02:04 zecke, probably the trouble is that oe built gpe-bluetooth this nite, while the libmimedir was the 0.2, while this morning I did an upgrade and libmimdir has been changed to 0.3 Nov 04 07:02:28 does somebody have somthing like a dataflow diagram that shows what happens? Nov 04 07:02:39 Or a sequence diagram? Nov 04 07:05:02 Cobelius: well do_configure and do_compile are shell functions Nov 04 07:05:15 Cobelius: consult your POSIX.3 documentation for scoping of shell Nov 04 07:05:32 Cobelius: for variables and such consult the bitbake documentation Nov 04 07:05:45 Cobelius: e.g most .bb files have a SRC_URI = "moo" Nov 04 07:05:53 Cobelius: you could do export SRC_URI = "moo" Nov 04 07:06:02 Cobelius: so it will be exported to the shell as well Nov 04 07:17:23 I currentyl trying to make a packag for the EPICS base Nov 04 07:18:10 in the logfile in the work directory stand "Nothing to compile" Nov 04 07:19:34 my .bb file is Nov 04 07:19:42 DESCRIPTION = "A scientific software" Nov 04 07:19:43 HOMEPAGE = "http://www.aps.anl.gov/epics/ Nov 04 07:19:43 LICENSE = "GPL" Nov 04 07:19:43 SRC_URI = "http://www.aps.anl.gov/epics/download/base/baseR3.14.7.tar.gz" Nov 04 07:19:43 DEPENDS = "readline" Nov 04 07:19:43 do_configure() { Nov 04 07:19:45 could you put that in pastebin.com? Nov 04 07:19:45 export EPICS_HOST_ARCH=linux-x86 Nov 04 07:19:47 } Nov 04 07:21:07 koen: the copy of the file? Nov 04 07:21:12 yes Nov 04 07:21:17 Okay Nov 04 07:21:59 http://pastebin.com/417159 Nov 04 07:22:33 I don't think that inherit base is needed Nov 04 07:23:10 what's the name of the .bb? Nov 04 07:23:16 base_3.14.7.bb? Nov 04 07:23:32 epics-base_3.14.7.bb Nov 04 07:23:43 I see Nov 04 07:24:11 S = "${WORKDIR}/base-${PV}" Nov 04 07:25:02 OE tried to do 'cd ${PN}-${PV}', but epics-base-3.14.7 doesn't exist Nov 04 07:25:51 see http://ewi546.ewi.utwente.nl/tmp/d/keys.html for an overview of keys Nov 04 07:26:47 ahh thank you Nov 04 07:30:35 ~lart Trolltech... Nov 04 07:30:35 * ibot resizes Trolltech...'s terminal to 40x24 Nov 04 07:52:24 zecke: stop bashing trolltech! Nov 04 07:54:56 re Nov 04 07:56:48 hey kergoth & Bernardo Nov 04 07:57:16 koen: http://www.trolltech.com/products/qtopia/indepth.html Nov 04 07:57:24 koen: "A Bluetooth and IrDA infrared beaming implementation ensures Qtopia can communicate with other mobile devices, including laptops. It is based on the OBEX standard." Nov 04 07:57:45 well yes they use OpenOBEX but they do not have a single line of bluetooth support in Qtopia itself Nov 04 07:58:06 http://www.trolltech.com/products/qtopia/pda.html Nov 04 07:58:11 Additionally, third party developers can create additional applications for the platform. To date, there are over 1,000 applications developed for Qtopia PDA. Nov 04 07:58:26 I thought TT didn't want us coding Software for non GPL platforms.. Nov 04 07:59:03 imagine! Nov 04 07:59:12 free development for a company! Nov 04 07:59:14 eeeeevil Nov 04 07:59:34 koen: you read the N770 on opie-devel? Nov 04 07:59:42 yes Nov 04 08:00:24 KDE and Qtopia are blowing there future(s)... Nov 04 08:00:33 I couldn't resist quoting some parts of it in #maemo Nov 04 08:00:40 we had a nice laugh Nov 04 08:01:00 TT employee saying 'don't use QT' Nov 04 08:01:02 priceless Nov 04 08:01:18 well differentiate a bit more Nov 04 08:01:34 it was funny nevertheless Nov 04 08:04:19 qtopia launcher is beatiful and so useful.. doctab is great.. Nov 04 08:04:20 argh. Nov 04 08:04:26 ~lart ljp Nov 04 08:04:26 * ibot duct-tapes ljp to the floor and drools on him Nov 04 08:16:44 koen: the biggest annoyance of Qtopia code is, you have the mixing of tabs and spaces Nov 04 08:16:48 even within lines... Nov 04 08:17:40 does 'indent' support c++? Nov 04 08:18:01 I imagine kdevelop has some indenter too Nov 04 08:18:53 koen: if you need to patch stuff in Qtopia code - which happens fairly often Nov 04 08:19:04 koen: you do not want to have too many white space changes... Nov 04 08:19:42 :set ts=8... well Nov 04 08:19:57 koen: indent does not support c++ Nov 04 08:20:51 zecke: ping Nov 04 08:21:25 indent will make a mess of c++ Nov 04 08:21:41 the closest I have found for C++ formating is astyle Nov 04 08:21:45 hey Nov 04 08:22:00 mithro: I have been going directly to bed after uni+work this week .( Nov 04 08:22:05 I'm going to get the binaries Nov 04 08:22:40 Crofton: indent can even fsck c++ code to be not compilable Nov 04 08:23:21 * koen remember running indent on php code Nov 04 08:23:37 I use (g)vim to indent Nov 04 08:24:44 zecke: okay - will be intrested to see if the binaries work Nov 04 08:26:08 mithro: they better do... Nov 04 08:26:28 zecke: you learnt a lesson yet? :P Nov 04 08:26:47 zecke: any chance you'll have time on the weekend? Nov 04 08:27:05 mithro: yes, it is scheduled for assignments though Nov 04 08:27:42 and starting my next DVB-? application... Nov 04 08:49:26 ouch Qt4 does not differentiate between signed and unsigned char.. where it should do so Nov 04 08:51:49 morning Nov 04 09:08:38 http://www.livejournal.com/users/mjg59/52975.html ;) Nov 04 09:10:00 heh Nov 04 09:10:38 hrw|work: who is mjg? Nov 04 09:10:41 hah Nov 04 09:10:51 zecke: dont know - found that on debian planet Nov 04 09:11:00 He's Matthew Garrett. Nov 04 09:11:07 It amuses me how quickly ljp was swung over to completely TT views Nov 04 09:11:08 Worked on Dasher. Nov 04 09:11:27 Does Linux laptop stuff. Generally a good guy. Excellent at ranting. Nov 04 09:12:05 Noodles: well we have Opie dasher; takes about 10 secs to type bitbake opie-dasher ;) Nov 04 09:12:51 kergoth: ever heard about MapFS? Nov 04 09:13:24 nope, what is it? Nov 04 09:13:29 kind of unionfs Nov 04 09:13:35 http://trends.newsforge.com/article.pl?sid=05/11/03/2255215 Nov 04 09:14:36 ok. I have to go - cu Nov 04 09:15:00 interesting. maybe it doesnt kernel panic as much as unionfs ;) Nov 04 09:15:10 zecke: Who did the initial work for the opie dasher build though? Nov 04 09:16:26 Noodles: dunno :( maybe it was mjg? Nov 04 09:17:11 I think he did some of the work, yes. He's not talking about it in terms of the 770 on that blog post. Nov 04 09:17:12 oh right Nov 04 09:17:13 well Nov 04 09:21:58 http://jacobo.tarrio.org/albums/cosas/gentu.png Nov 04 09:28:12 mithro: hehe FBSD 6.0 is out, time to upgrade (by self compiling) Nov 04 09:29:33 zecke: re: bitbake fetcher separation, http://www.rpsys.net/openzaurus/temp/bitbake_modularise_fetcher-r0.patch is what I'm currently working from Nov 04 09:32:56 RP: looks good. I'll make it autoload the stuff if you're finished Nov 04 09:33:30 mickeyl: please feel free Nov 04 09:33:43 s/if/when/ of course :) Nov 04 09:34:04 mickeyl: the unfinished Bk one should probably get wiped and the rest of the code can probably be cleaned up a bit Nov 04 09:34:31 mickeyl: but I thought first off, it was just best to split them "as existing" Nov 04 09:34:37 yep Nov 04 09:34:39 RP, I haven't time to discuss the bug right now, I whish I'll have more time in the evening or in the next few days Nov 04 09:34:45 (although I did add a git parser) Nov 04 09:34:56 s/parser/fetcher/ Nov 04 09:35:19 Pigi: No problem, we'll find the right time eventually - its not urgent Nov 04 09:35:29 ok, thx Nov 04 09:36:17 mickeyl: That patch should probably go into svn as is, then we can make further alterations after that. Nov 04 09:36:49 ok, i'll take a go at testing at this weekend Nov 04 09:37:16 mickeyl: I'll bugzilla it? Nov 04 09:37:22 RP: yes, please Nov 04 09:38:59 mickeyl: I'll admit to being clueless when it comes to sorting out the API for it properly in python so someone with better knowledge of python will have to do that ;-) Nov 04 09:39:32 mickeyl: autoload will even work with a installed version of bitbake? Nov 04 09:39:46 mickeyl: what is your opinion on del usage Nov 04 09:43:29 hey mickeyl Nov 04 09:44:31 RP: the svn fetcher needs work as well. it was clearly done by someone used to working with cvs, rather than someone who knows svn Nov 04 09:44:32 heh Nov 04 09:46:38 kergoth: I suspect they all need some TLC :). Even the git one isn't ideal at the moment... Nov 04 09:49:20 zecke: del ? Nov 04 09:49:35 zecke: in what place? or general? Nov 04 09:49:46 mickeyl: in general (python) Nov 04 09:50:34 zecke: ah. I use del rarely. Nov 04 09:50:50 mickeyl: Does it make a difference to python's memory consumption? Nov 04 09:51:12 RP: that depends whether the reference you remove is the last one to a known object or not Nov 04 09:51:17 del doesn't delete an object Nov 04 09:51:19 03koen 07org.oe.dev * r0903fa8c... 10/: xserver-kdrive 20050207: make simpad use the apm hack too Nov 04 09:51:20 del removes a reference Nov 04 09:51:30 and only if this reference was the last one, the obj gets deleted Nov 04 09:53:03 hmm.. the reference is automatically removed when the var goes out of scope, isnt it? Nov 04 09:53:08 mickeyl: and if you go out of scope one line later Nov 04 09:53:13 heh Nov 04 09:53:14 later all Nov 04 09:53:26 mickeyl: del hurts more than it helps Nov 04 09:54:11 like i said, i use it rarely. btw., if it's the last reference, there's nothing to go out of scope afterwards Nov 04 09:54:25 err no Nov 04 09:54:37 s/if it's the last reference// Nov 04 09:55:40 mickeyl: There are some dubious looking del statements in the fetch modules then... Nov 04 09:55:58 i'll glance over 'em Nov 04 09:58:04 bbl Nov 04 10:03:35 hi all Nov 04 10:05:00 * CosmicPenguin hasn't updated in about 3 weeks Nov 04 10:05:03 this is going to take a while Nov 04 10:06:14 what is ? Nov 04 10:08:40 S1nn3r: the montone update Nov 04 10:08:56 ~ewi Nov 04 10:08:57 methinks ewi is ewi546.ewi.utwente.nl - main backup server for OE monotone. Hosts database snapshot: http://ewi546.ewi.utwente.nl/OE/OE.db.bz2 Nov 04 10:09:23 CosmicPenguin: unless you have private branches the snapshot is your best bet Nov 04 10:11:08 of course with you 64way 20GHz opterons monotone is quite fast ;) Nov 04 10:11:13 your* Nov 04 10:16:05 Hey, I'm cool Nov 04 10:16:09 its not too bad Nov 04 11:00:34 re Nov 04 11:01:58 how big is the rootflash size on poodle? Nov 04 11:02:15 the .bb says 32, but I'm inclined to not believe that Nov 04 11:02:46 koen: I think that is right Nov 04 11:02:53 32 flash + 32 ram Nov 04 11:03:17 so it's wrong Nov 04 11:03:28 rootflash size = size of / Nov 04 11:03:57 the c7x0.conf is wrong too Nov 04 11:04:13 my corgi only has ±25 MB for /, not 32MB Nov 04 11:04:17 * koen files a bug Nov 04 11:06:21 so rootflash is the flash size - kernel partition size? Nov 04 11:09:43 size of the mtd partition that holds / Nov 04 11:10:04 i.e. 'how big a rootfs can be' Nov 04 11:13:31 so rootflash sets size of / flash area? Nov 04 11:14:16 it indicates it Nov 04 11:14:38 can you set things up so say like /usr is mounted elsewhere? Nov 04 11:14:48 like if you have a 4G compact falsh drive? Nov 04 11:14:52 no, it's purely informational Nov 04 11:14:59 bummer Nov 04 11:15:16 I need to build something that won't fit in flash Nov 04 11:15:20 opie and gpe image use it to include/exclude packages Nov 04 11:15:38 looking for a way to automate the file system image generation Nov 04 12:44:48 hey Nov 04 12:46:28 zecke: hey Nov 04 12:48:58 RP: thanks for posting the patch Nov 04 13:26:47 hi Nov 04 13:27:01 hey Nov 04 13:27:08 hey hrw Nov 04 13:30:00 hmm.. petitepainture 1.5 does not build Nov 04 13:30:50 missing patch. Nov 04 13:32:39 hi hrw Nov 04 13:35:15 ok. bug added and assigned Nov 04 13:43:50 witch was the target for pxa250 kernel for sl-5600 Nov 04 13:43:54 which Nov 04 13:44:31 darkschneider: virtual/kernel :) Nov 04 13:44:49 darkschneider: only pxa255 poodle differ - default poodle kernel is for pxa250 Nov 04 13:46:46 blah Nov 04 13:49:56 * darkschneider mumbles and feels lost like allways Nov 04 13:50:18 03koen 07org.oe.dev * ra2e7ad19... 10/: gpe-sketchbook: add 0.2.9 Nov 04 13:57:38 btw - one of my OE patches was finally applied upstream (gammu one) Nov 04 13:57:57 'night Nov 04 13:58:08 night Bernardo Nov 04 14:06:44 03hrw 07org.oe.dev * rcfd619e5... 10/: openttd: upgraded to 0.4.0.1 Nov 04 14:08:06 PV="0.2.0-oesf0.2.4" (corsair) looks weird Nov 04 14:08:29 any suggestions how to PV it? Nov 04 14:08:55 blech Nov 04 14:08:59 upstream is 0.2.0, one guy from oesf forum wrote some patches and bumped version to 0.2.4 Nov 04 14:09:02 keep that herecy outside OE Nov 04 14:09:15 name-stealing bastards Nov 04 14:09:17 patches was mainly bugfixes Nov 04 14:09:33 so suggest better PV :) Nov 04 14:09:46 0.2.0-0.2.4? Nov 04 14:09:48 0.2.0.4? Nov 04 14:10:00 0.2.0.4 sounds ok Nov 04 14:10:11 upstream author does not react yet (got patches) Nov 04 14:10:18 or just r4 Nov 04 14:10:25 since it are only patches Nov 04 14:11:40 ok Nov 04 14:12:14 ~lart corsair 'maintainer' Nov 04 14:12:15 * ibot squeezes corsair 'maintainer' till corsair 'maintainer' turns blue like papa smurf Nov 04 14:12:41 I will probably remove maintainer field or takeover it Nov 04 14:24:31 I'm using 2.6.14-git3 kernel but the modules are in /usr/lib/2.6.14-rc4 Nov 04 14:24:33 is it normal? Nov 04 14:24:39 what could be the problem? Nov 04 14:25:01 katossi: rc4 are never then git3 for ipkg Nov 04 14:25:14 never? Nov 04 14:25:18 newer? Nov 04 14:25:19 newer Nov 04 14:25:42 edit ipkg status file and change 2.6.14-rc4 to something older (2.6.13 for example), then upgrade modules Nov 04 14:25:48 ok, I think I know what I have to change Nov 04 14:29:39 ~lart wifi Nov 04 14:29:39 * ibot moos at wifi Nov 04 14:30:37 I 'love' blinking led on my wifi card.. Nov 04 14:30:57 heh Nov 04 14:31:57 djeez Nov 04 14:32:05 'ipkg -h' is too hard for pdaX morons Nov 04 14:32:21 ;how to remove orphans; thread? Nov 04 14:32:24 hrw: i love it too, my card just blinked from the day i bougth it to today :) and it never ever sent a packet :) Nov 04 14:32:49 darkschneider: if you not needed it I can take it ;) Nov 04 14:33:26 hehe, ehy it's a offiacly supported wifi card, can worth about 10 dollars Nov 04 14:33:40 shippung it to you is probably more expensive Nov 04 14:33:43 which one is it? Nov 04 14:34:01 pretec 802.11b Nov 04 14:34:18 it's orinoco driver if i rember Nov 04 14:34:20 hrw: do you it's rude to paste the output of ipkg -h ? Nov 04 14:34:22 a.. they were <30$ at hsn.com Nov 04 14:34:35 koen: you are known are rude already Nov 04 14:35:06 hrw yes, good card consume a few and best supported in linux, the only fact is that i never managed to make it work Nov 04 14:35:14 ~spell archaic Nov 04 14:35:16 'archaic' may be spelled correctly Nov 04 14:36:02 hmmm, lardman isn't helping by saying oz uses a rewritten ipkg Nov 04 14:36:23 correct him too Nov 04 14:36:36 strictly speaking it is try Nov 04 14:36:37 CIao all Nov 04 14:36:39 true* Nov 04 14:36:42 hey Pigi Nov 04 14:36:54 hey koen Nov 04 14:37:00 but it was rewritten by the ipkg people Nov 04 14:37:25 pdaX use ipkg 0.99.xxx Nov 04 14:37:47 well, nice to know someone other use ipkg :) Nov 04 14:37:50 hi hrw Nov 04 14:39:54 ops.. sorry... Nov 04 14:40:08 its shell one... Nov 04 14:40:31 atleast that one in pdaxbuilder Nov 04 14:40:37 # modified by sharp 2001 Nov 04 14:40:40 # modified by cacko.biz - pdaXrom.org 2003, 2004 Nov 04 14:40:58 hrw, I've seen you did the port for kdepim under X11. Is that true ? Nov 04 14:41:57 Pigi: its not port as kdepimpi can be used under x11 Nov 04 14:42:08 I started work on building kdepimpi/x11 in oe Nov 04 14:42:16 http://bugs.openembedded.org/show_bug.cgi?id=371 Nov 04 14:42:27 I have a question for you if you have some cpu cycle free for me Nov 04 14:42:34 go Nov 04 14:42:59 I've managed to build it, but it seems it doesn't work . Did you tried it ? Nov 04 14:43:42 no Nov 04 14:44:04 ok. I'll spent some time on it this w.e. (if I found some free time). Nov 04 14:44:25 BTW, basically it starts, but seems to hang forevere eating 100% cpu. Nov 04 14:45:15 could you try stracing it? Nov 04 14:45:25 obviuosly :) Nov 04 14:45:32 brk(0x325000) = 0x325000 Nov 04 14:45:32 brk(0x349000) = 0x349000 Nov 04 14:45:32 mmap2(NULL, 135168, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x41297000 Nov 04 14:46:37 that's the lasts strace output before it simply stand here without doing anything. Nov 04 14:47:11 any suspicious NOENTs above that? Nov 04 14:47:39 dumoing the output to file for checking it. Nov 04 14:47:44 dumping even Nov 04 14:49:52 03hrw 07org.oe.dev * rbccb4d45... 10/: Nov 04 14:49:53 corsair: package takeover, cleaning, patching Nov 04 14:49:53 - added corsair-024.patch with bugfixes from 'mastermix' Nov 04 14:49:53 details: http://www.oesf.org/forums/index.php?showtopic=14661 Nov 04 14:49:53 - removed two pictures which collide with opie-pics and (R)DEPENDS on it Nov 04 14:49:53 - cleaned DESCRIPTION Nov 04 14:50:48 cu Nov 04 14:50:51 basically the pics seems to be missing ( the directory it points is wrong Nov 04 14:51:11 does the x11 version still use those ugly /opt paths? Nov 04 14:51:42 yes Nov 04 14:51:54 /opt/QtPalmtop Nov 04 14:53:02 ugh Nov 04 14:53:11 the x11 version should use /usr Nov 04 14:53:17 maybe this could be suspect: Nov 04 14:53:19 open("/usr/X11R6/lib/X11/icons/default/cursors/xterm", O_RDONLY) = -1 ENOENT (No such file or directory) Nov 04 14:53:20 open("/usr/X11R6/lib/X11/icons/default/index.theme", O_RDONLY) = -1 ENOENT (No such file or directory) Nov 04 14:53:25 and the binary should be named gpe-calendar ;) Nov 04 14:53:51 koen: monotone build using fink ;) Nov 04 14:57:20 kergoth: how was that site with free svn services called? Nov 04 15:03:52 later all Nov 04 15:04:16 stupid gdm Nov 04 15:04:18 * CosmicPenguin resets Nov 04 15:38:44 I'm making a bitbake file for a new program and I'm having a little difficulty, is there anyone around that can spare a minute? Nov 04 15:39:41 try Nov 04 15:41:19 okay, I'm making a .bb for void11. So I need to download the source for it (no problem) and the source for hostapd-0.1.3 (here is the problem). Both need to be unpacked, and void11 just needs the unpacked source of hostap-0.1.3 to compile... however, bitbake seems to think that hostapd-0.1.3 is a patch and it fails at that point. So... what do I do? Nov 04 15:42:39 I can post any needed output of my .bb on pastebin if needed Nov 04 15:46:07 Zero_Chaos: why does it need hostap-0.1.3 source? Nov 04 15:46:35 Zero_Chaos: basicly your SRC_URI will have http://void13.bz2 and http://hostap-stuff.tar.bz2 in it Nov 04 15:46:44 Zero_Chaos: feel free to pastebin Nov 04 15:46:47 zecke: it needs it to compile.... I have no idea, I didn't write it. must be some header files Nov 04 15:46:59 zecke: I'm pastebining now... one sec Nov 04 15:47:16 Zero_Chaos: well if it needs a header you should create two bb files Nov 04 15:47:25 Zero_Chaos: one managaing hostap and staging of headers Nov 04 15:47:40 Zero_Chaos: and void... should DEPENDS on it Nov 04 15:47:58 zecke: that isn't a bad idea at all, I'll try that and if I fail I'll pastbin that failure Nov 04 15:48:05 zecke: thanks for your help, I think I've got it now. Nov 04 15:51:17 zecke: if I don't want to configure or compile or anything, I just want it to unpack and exist.... what do I do? just leave out everything but SRC-URI? Nov 04 15:52:18 Zero_Chaos: yes, or create 'empty' do_compile() {} and do_configure() {} lines Nov 04 15:52:27 empty as in, only containing ':' Nov 04 15:52:45 can I just leave that all out? Nov 04 15:53:26 try it ;) Nov 04 15:53:40 it takes so long to parse :-( Nov 04 15:53:45 here goes Nov 04 15:53:51 Zero_Chaos: bitbake -b Nov 04 15:54:00 Zero_Chaos: you can only bake one file Nov 04 15:54:09 thanks Nov 04 15:54:10 Zero_Chaos: or use bitbake -i to drop into the interactive mode Nov 04 15:54:21 bitbake -b foo.bb Nov 04 15:54:26 bitbake -b foo.bb -c compile -f Nov 04 15:54:45 to force bitbake to execute the compile step of foo.bb Nov 04 15:56:06 where do I put in a version number? it defaulted to 1.0 Nov 04 15:56:17 just file_version.bb? Nov 04 15:56:24 Zero_Chaos: yes Nov 04 15:56:40 thanks' Nov 04 15:56:40 or set PV = "3" Nov 04 15:57:06 bitbake tries to make PackageVersion (PV) and PackageRelease (PR) up from the filename Nov 04 15:57:12 but you can override it Nov 04 15:59:43 okay, here we go... it looks good so I touched local.conf and I'm going to let it work normally and see what happens Nov 04 16:00:14 once I get this to work and test it (along with my other .bb files) where do I submit to be added? Nov 04 16:01:05 Zero_Chaos: mailing list or better bugs.openembedded.org Nov 04 16:01:18 okay Nov 04 16:01:20 thanks Nov 04 16:01:33 you are welcome Nov 04 16:04:40 ugh, I hate watching it parse.. I need a faster...something/everythign Nov 04 16:05:02 bitbake -b ? Nov 04 16:05:23 koen: I know, but I wanted to make sure it worked without doing that. Nov 04 16:05:47 in that case you might like bitbake -i Nov 04 16:06:22 koen: Anyway I think mono is declining Nov 04 16:06:44 zecke: and we are still not filthy rich Nov 04 16:06:51 koen: I don't know much about any of this bitbake stuff, I can run it and I pretty much leave it at that, I don't use it enough to justify learning it well. Nov 04 16:21:53 is there a way to make bitbake parse a file and tell hm what is wrong? Nov 04 16:35:31 openembedded.org is down? Nov 04 16:35:46 yes, use oe.handhelds.org instead Nov 04 16:35:50 oki Nov 04 16:35:52 * koen looks at kergoth Nov 04 17:11:11 away Nov 04 17:11:15 doh Nov 04 17:11:25 how do I make an empty patch section? Nov 04 17:11:32 do_patch() {} Nov 04 17:11:33 ? Nov 04 17:13:55 I don't need to patch anything, yet it keeps failing on do_patch. How do I make it acknoledge that there is no patching and it should just succeed? Nov 04 17:19:16 I don't need to patch or configure... but it fails... can anyone help me? Nov 04 17:23:45 Zero_Chaos: well, please paste the 'patch error' Nov 04 17:23:52 and your bb file Nov 04 17:24:04 it will only apply a patch if you have something like Nov 04 17:24:13 file://moo.foo;patch=1 Nov 04 17:26:24 zecke: pastebin.ca/27663 is both of my .bb files (the one for the headers and the main program). The headers work fine, just as I want, but the other one is a MESS and I have no idea what to do. Nov 04 17:27:06 zecke: I'm trying to pass some options to make as you can tell, and I have no idea what I'm doing. Nov 04 17:27:25 zecke: and like I said, it fails on do_patch, then on do_configure, then on do_compile Nov 04 17:29:02 brb Nov 04 17:31:02 sorry I'm too tired to think three thoughts straight Nov 04 17:31:14 you would need to paste the error as well Nov 04 17:49:03 zecke: pastebin.ca/27671 Nov 04 17:52:55 * CosmicPenguin hugs bvi Nov 04 17:53:07 Zero_Chaos: considered looking into the script that failed? Nov 04 17:57:21 it's do_patch, and I'm not trying to patch Nov 04 17:59:00 got to go, sorry Nov 04 17:59:01 thanks Nov 04 20:51:41 zecke: still awake by any chance? Nov 04 23:27:58 RP: Do you have documentation on the SCOOP chip or device? Nov 05 00:10:02 hey all! Nov 05 02:05:22 xiaowen: Only what you read in the driver or is in my head... Nov 05 02:06:02 RP: So you wrote the 2.6 driver based entirely on what you read in 2.4? Nov 05 02:06:32 xiaowen: yes Nov 05 02:06:51 xiaowen: Although I do now have a fairly good idea of what the thing does... Nov 05 02:07:12 Hehe. Nov 05 02:07:52 I'm trying to see how LEDs get implemented. In 2.4, it looks like they just write to 0xf2200064. Nov 05 02:08:19 The scoop chips have gpios. You write to the scoop gpio register to turn them on and off Nov 05 02:10:25 OK, my questions might sound stupid, sorry .. Nov 05 02:10:45 So you have GPIOs in the PXA255, and those are connected to the SCOOP somehow? Nov 05 02:11:11 no, the pxa gpios are totally seperate from the extra ones on scoop Nov 05 02:11:44 scoop is more than just the gpios - its a buffer chip for the CF interface which just happens to provide the gpios as an extra Nov 05 02:12:07 And this buffer chip also controls the LEDs? Nov 05 02:12:16 On some devices Nov 05 02:12:45 Is this buffer chip memory mapped on the CPU, so my job is to find what the memory address is? Nov 05 02:13:02 The 2.6 driver tells you all this Nov 05 02:13:17 Where? Nov 05 02:13:29 For which device? Nov 05 02:13:33 Are you wanting to play under 2.6 or 2.4? Nov 05 02:13:38 2.6. Nov 05 02:13:43 which machine? Nov 05 02:13:46 tosa Nov 05 02:17:20 set_scoop_gpio(&tosascoop_jc_device.dev, TOSA_SCOOP_JC_NOTE_LED) Nov 05 02:17:20 reset_scoop_gpio(&tosascoop_jc_device.dev, TOSA_SCOOP_JC_NOTE_LED) Nov 05 02:17:37 There's a nice big hint about what you're trying to do ;-) Nov 05 02:18:15 I'm sure I have no idea what you mean, but thank you. :) Nov 05 02:18:23 lookup some of those names if you want to see how it works Nov 05 02:18:47 Those two lines of code will turn the led on and then off again Nov 05 02:19:35 Did you just write that or is that in the source? I don't see it. Nov 05 02:20:27 Look for set_scoop_gpio which is in the scoop driver (scoop.c). Also look for the scoop device declared in tosa.c. The LEDs are declared in the tosa machine header file tosa.h Nov 05 02:20:38 Combine them and you can turn the LEDs on and off... Nov 05 02:21:07 tosa probably doesn't do led support yet so there is no code matching the above exactly... Nov 05 02:21:39 I have other things I need to do so you'll have to figure anything else out for yourself I'm afraid Nov 05 02:22:15 OK, while I go investigate (bang my head) some more, can you please explain arch/arm/mach-pxa/generic.c:standard_io_desc? What's the physical address? Nov 05 02:22:41 (Yes, according to the status page, it doesn't, that's why I'm trying. -_-) Nov 05 02:23:41 I looked at the memory map diagram in the PXA255 developer's manual, and I guess that lists virtual addresses, so where is the physical address coming from? Nov 05 02:24:11 dudes: How can I "package" something? Nov 05 02:24:37 hrw made kdepimpi-x11 compile, but lost interest / has no time, so I want to package his work. Nov 05 02:24:57 is there a good instruction set somewhere close by for that? Nov 05 02:25:38 I've no idea sorry. I (re)started looking at this yesterday. -_- Nov 05 02:35:34 * xiaowen goes back to reading LDD3/Device Model. -_- Nov 05 02:44:21 xiaowen: You have physical addresses which are where things are really located and virutal ones which is the address you write to as a device driver. The MMU translates between the two. Nov 05 02:44:55 SCOOP used to have a fixed mapping under 2.4. Under 2.6, it gets remapped with ioremap (in scoop.c) Nov 05 02:45:21 However you don't really need to know all this Nov 05 02:51:49 LDD talked about ioremap ... but I think I never understood where the physical address came from. They gave an example of writing to a parallel port, but I didn't get how they knew what the address of the parallel port is ... Nov 05 02:55:30 xiaowen: Its a feature of the hardware - where the physical chips address lines are wired to Nov 05 02:55:54 tosa.c declares the physical addresses of its scoop chips Nov 05 02:58:46 Zero_Chaos: sorry no **** ENDING LOGGING AT Sat Nov 05 02:59:57 2005