**** BEGIN LOGGING AT Tue Jul 05 02:59:57 2011 Jul 05 06:13:06 Anyone has experience with dhcpd and ad-hoc networks? Jul 05 06:13:34 what do you need davidlt ? Jul 05 06:14:00 I moved from Master to ad-hoc mode (libertas driver) and I have dhcpd Jul 05 06:14:25 On my machine I have eth0 and eth1 on 192.168.X.X, while wlan0 should be on 10.2.2.X Jul 05 06:14:42 I have dhcpd to deliver IP addresses via wlan0 Jul 05 06:14:51 It worked very well on Master mode Jul 05 06:15:04 Now I switch to ad-hoc and it doesn't work any more. Jul 05 06:15:26 dhcpd detects machine and offers lease (IP), but my iPhone or Mac doesn't take it. Jul 05 06:15:30 dhcpd keeps offering it. Jul 05 06:15:57 While MacBook and iPhone request for renew lease for old IP with different subnet Jul 05 06:16:00 guess the Mac/iPhone can't handle that...sounds like the driver/dhcpd are working fine. Jul 05 06:16:22 ah...again...sounds like its a mac side issue. Jul 05 06:16:53 2011-07-05T12:12:46.000000+02:00 overo dhcpd: DHCPREQUEST for 192.168.1.65 from 00:1b:63:04:29:ea via wlan0: ignored (not authoritative). Jul 05 06:17:11 I think same goes for Windows Jul 05 06:17:24 2011-07-05T12:12:55.000000+02:00 overo dhcpd: DHCPOFFER on 10.2.2.21 to 00:1b:63:04:29:ea (david-mac) via wlan0 Jul 05 06:17:24 2011-07-05T12:12:56.000000+02:00 overo dhcpd: DHCPDISCOVER from 00:1b:63:04:29:ea (david-mac) via wlan0 Jul 05 06:17:34 right..dhcpd says "I don't have authority to give out IP addresses for that range" Jul 05 06:17:44 192.168.1.65 is david-mac IP address from my local router Jul 05 06:18:23 dhcpd is working fine. Jul 05 06:18:30 I had manage to make it work before, but I can't reproduce it anymore Jul 05 06:18:42 But all the changes was made on my Linux machine Jul 05 06:18:51 what device was doing the offering and ignoring the requests? Jul 05 06:19:33 Gumstix Overo Fire Jul 05 06:20:14 I also installed olsrd package, but I don't fully sure that is needed Jul 05 06:20:48 it sounds like you have 2 competing dhcpd servers...your "local" router and this overo Jul 05 06:21:10 unless you are using 2 different interfaces on the mac. Jul 05 06:21:31 Yea, it's like all machines would think, that they are renewing lease on my local router Jul 05 06:22:07 At the moment Ethernet for internet and WiFi for connecting to Gumstix Jul 05 06:23:04 okay using wifi you hit the renew on the preference panel for Airport and it doesn't work? Jul 05 06:23:13 and thats what you record? Jul 05 06:23:23 Yeah Jul 05 06:23:40 3 ignores and continuous DICOVER/OFFER Jul 05 06:24:25 might be best to read up on diffences between infrastructure and ad-hoc for dhcp things. Jul 05 06:25:05 Tried googling, but ad-hoc mode for dhcpd, which is not supported any more. So it's hard to get some good reading. Jul 05 06:25:06 it sounds like without an AP that the mac can't deal with it correctly. Jul 05 06:25:41 since the behaviour is coming not from the overo but rather the mac. Jul 05 06:26:28 the overo has control over 10.2.2.X and the mac is requesting DHCP like its on en0 Jul 05 06:26:53 this really smells like a Mac issue. Jul 05 06:27:33 Hmm.. I think Windows works similarly Jul 05 06:27:39 Could test in few minutes Jul 05 06:28:22 Local IP for wlan0 is also fetched from dhcpd and it works. Jul 05 06:28:28 Let me test with Windows Jul 05 06:29:25 The strange thing is that I had it working just my playing with avahi, olsrd, dhcpd and creating network. Jul 05 06:29:44 But I can't reproduce the steps and re-flashing NAND. Jul 05 06:35:59 ka6sox-home: same for Windows, it doesn't take IP from DHCP Jul 05 06:36:27 So it uses self-assigned IP Jul 05 06:38:00 ka6sox-home: it worked after I restarted networking Jul 05 06:39:05 I also get: libertas: EVENT: unknown event id 30 Jul 05 06:39:09 And WEP doesn't work Jul 05 06:46:54 davidlt, still sounds like the dhcp issue is the mac/iphone not correctly dealing with it. Jul 05 06:47:17 the WEP etc is connectivity...not things like dhcp (which comes after) Jul 05 06:47:28 Windows also doesn't accept IP from dhcpd Jul 05 06:47:58 Yeah, just noticed that if I set iwconfig wlan0 key <...>, it is still open Jul 05 06:48:02 is it complete? Jul 05 06:48:10 The key? Jul 05 06:48:20 aka, ip,netmask,gateway, dns servers? Jul 05 06:48:54 Sorry, didn't understand Jul 05 07:03:16 hi all Jul 05 07:05:17 ka6sox-home: This is a sequence of commands that does work: http://pastebin.com/ChmqJdpg Jul 05 07:06:30 interesting. Jul 05 07:06:40 the last dhclient is also interesting. Jul 05 07:07:07 Why? It's not required, just used it to see that dhcpd it working Jul 05 07:07:24 It returns static IP for the Gumstix Jul 05 07:07:59 Encryption key:off Jul 05 07:08:04 Power Management:on Jul 05 07:08:32 But I did set the key and another strange thing is that libertas doesn't support PSM, but shows as on. Jul 05 07:08:48 davidlt, so you have setup reservation? Jul 05 07:09:07 What do you mean by setup reservation? Jul 05 07:09:51 dhcpd can give IP addresses out based on mac addresses. Jul 05 07:10:21 I hard coded one IP address, that's for Gumstix itself Jul 05 07:10:38 kk Jul 05 07:10:51 thats why I asked why the dhclient... Jul 05 07:13:25 Okay, I can't turn off the power saving mode, but maybe that's requirement in ad-hoc mode Jul 05 07:13:42 iwconfig wlan0 key 1111-1111-1111-1111 doesn't work Jul 05 07:13:57 But iwconfig wlan0 key 1234567890 , set the key: Encryption key:1234-5678-90 Jul 05 07:14:05 hm? Jul 05 07:14:16 whats your kernel version? Jul 05 07:14:20 .39 Jul 05 07:14:27 use iw Jul 05 07:14:30 wpasupplicant Jul 05 07:14:32 and ifup Jul 05 07:14:55 and use mac80211 Jul 05 07:16:45 and if I remember correctly Jul 05 07:16:52 woglinde: iw is not included in my distro, I tried wpa_supplicant yesterday trying to add WPA, but that failed Jul 05 07:17:00 you cannt set wpa stuff anyway with iwconfig Jul 05 07:17:10 what failed Jul 05 07:17:18 whats your wifi chipset? Jul 05 07:18:19 Wi2Wi based on Marvell's 88W8686 (driver: libertas) Jul 05 07:18:33 WEP stuff worked with libertas_tf (I think), but was super unstable Jul 05 07:18:42 Tons of errors from driver Jul 05 07:20:52 * davidlt restarting Gumstix to try wpa Jul 05 07:21:12 libertas: Timeout submitting command 0x0000 Jul 05 07:21:12 libertas: Received CMD_RESP with invalid sequence 46 (expected 0) Jul 05 07:21:14 Uch stuck Jul 05 07:25:29 ieehks Jul 05 07:25:34 libertas Jul 05 07:25:35 hm Jul 05 07:25:36 woglinde: output of wpa_supplicant: http://pastebin.com/3RCXWRTe Jul 05 07:25:58 I can see TestMe from my Mac, but still open Jul 05 07:26:31 does the driver support mac80211 Jul 05 07:26:38 instead of wirelessapi? Jul 05 07:27:11 hm Jul 05 07:27:12 http://wireless.kernel.org/en/users/Drivers/libertastf Jul 05 07:28:21 is this your host Jul 05 07:28:27 or embedded device? Jul 05 07:28:33 embedded device Jul 05 07:28:51 hm with libertas/libertas_tf from oe? Jul 05 07:28:55 I was using libertastf before, have the latest stuff Jul 05 07:29:14 did you configure wpa_supplicant to use mac80211? Jul 05 07:29:15 libetas_tf was very unstable so I decided to try libertas Jul 05 07:29:26 hm okay Jul 05 07:31:41 Don't know, just used simple wpa_configuration I found on web Jul 05 07:36:41 http://pastebin.ubuntu.com/638256/ Jul 05 07:36:47 I have on my debian host Jul 05 07:37:06 after wpa-psk the key Jul 05 07:37:31 getting from wpa_passphrase Jul 05 07:40:31 OMG firefox5 is eating 170MB only for opening Gmail + Chatzilla Jul 05 07:41:17 even 183MB now Jul 05 07:43:30 woglinde: looking, I think I have to update my wpa_supplicant Jul 05 07:49:27 Gonna build new OS image with iw and new wpa_supplicant Jul 05 07:52:11 hello ... I had compiled a alsa-lib and alsa-utils .....but when I trying to play .... aplay sample.wav .... I m getting following errors... Jul 05 07:52:19 http://pastebin.mozilla.org/1264803 Jul 05 07:53:52 rocker: so, this is not a machine support channel... but it looks like you have no sound device Jul 05 07:54:47 but I have given support in kernel... Jul 05 07:55:36 I'm no expert but don't you need to do some alsa setup as well? Jul 05 07:55:43 i.e. configuration Jul 05 08:47:19 What is requirements for new Bitbake? Jul 05 08:48:53 python 2.6 Jul 05 08:49:24 It crashes: AttributeError: 'NoneType' object has no attribute 'split' Jul 05 08:50:45 running 2.6.6 Jul 05 08:50:57 hm Jul 05 08:51:05 python -version? Jul 05 08:51:19 -- Jul 05 08:51:24 python2.6 --version Jul 05 08:51:25 Python 2.6.6 Jul 05 08:51:45 I also adjusted bitbake : #!/usr/bin/env python2.6 Jul 05 08:54:33 I cleaned all cache, let's see Jul 05 08:54:36 Still same Jul 05 08:59:35 hm strange Jul 05 08:59:57 Trying 1.13.1 Jul 05 08:59:58 Same Jul 05 09:00:36 sorry have no clue at the moment Jul 05 09:01:15 Going back to 1.12.0 Jul 05 09:01:23 Looks like working Jul 05 09:05:52 davidlt: are you using stable or current master of OE? Jul 05 09:07:20 I justed cloned git://git.openembedded.net/bitbake and checkout 1.13.2 Jul 05 09:08:06 no, I mean the OE metadata Jul 05 09:08:08 While OE metadata is from Gumstix OE branch, which is mostly identical, but it's older Jul 05 09:08:13 oh right Jul 05 09:08:14 lol Jul 05 09:08:17 sorry Jul 05 09:08:21 gusmtix breaks so often Jul 05 09:08:30 I hope it will be better with core Jul 05 09:08:37 well, in this case it's a fix that's in OE but obviously not the gumstix branch Jul 05 09:08:44 so the only have to care about bsp Jul 05 09:09:13 They didn't merge latest things for few months already, I think Jul 05 09:09:23 yeah, there ought to be a gumstix bsp layer for oe-core, that would help a lot Jul 05 09:09:58 davidlt: so you can stick with 1.12.x or go to 1.13.2 and cherry-pick 6f3334a9a1ef5953a0a9e2b6ec92705a3f9295d1 from OE on top of your gumstix checkout Jul 05 09:11:10 I am probably gonna stick with 1.12.0 as that does the job Jul 05 09:11:21 Any significant changes in 1.13 ? Jul 05 09:12:31 Maybe you can even suggest alternatives for Gumstix? (wlan and eth is requirement, microSD support too, small size too) Jul 05 09:13:39 I'm not aware of anything to really match it sizewise Jul 05 09:15:10 I am used to the classical OE, how does the new structure work? Jul 05 09:16:34 davidlt do you need to develop? Jul 05 09:16:43 otherwise try narcissus Jul 05 09:17:55 We build custom omap3-console-image image, different configurations, modified packages, own written packages Jul 05 09:19:22 hm okay Jul 05 09:21:53 davidlt: re 1.13.2, quite a lot of changes, too many to list... if you really want to know, do a git log 1.12.0..1.13.2 :) Jul 05 09:22:49 davidlt: the new structure has oe-core as a base layer, then machine/distro/other layers on top Jul 05 09:23:33 classic OE tree is still updated? Jul 05 09:23:37 sort of Jul 05 09:24:05 some people are still working with it, but the idea is to move OE to be based on top of oe-core Jul 05 09:24:19 that means splitting out OE's metadata into various layers as appropriate Jul 05 09:24:29 davidlt not really Jul 05 09:24:36 better switch to oe-core Jul 05 09:24:40 and make your own layer Jul 05 09:25:04 yeah it really is a better model, especially if you want to track base OE but apply your own customisations as well Jul 05 09:25:07 Are where any details on that in wiki? Jul 05 09:25:14 no not yet Jul 05 09:26:45 I doubt that Gumstix is gonna use new structure for a long time. Jul 05 09:26:59 They haven't updated the stuff for 3 months Jul 05 09:27:43 I was using classical OE to grab some newer images and update by metadata Jul 05 09:28:45 and? Jul 05 09:28:51 make it yourself Jul 05 09:28:55 insnt that hard Jul 05 09:29:33 That would be probably nice, but I don't think that my boss would agree on that :) Jul 05 09:29:59 you have to raise it to your boss Jul 05 09:30:08 because oe-core is the future Jul 05 09:40:02 woglinde, where to find docs on oe-core ? Jul 05 09:43:56 sakoman is working on bsp for gumstix (koen said) Jul 05 09:44:01 configure: error: gmp.h can't be found, or is unusable. Jul 05 09:44:02 :P Jul 05 09:44:23 bsp layer for gumstix, that's what I've meant Jul 05 10:00:29 ynezz: cool Jul 05 10:01:04 woglinde: I used this example with wpa_supplicant: https://help.ubuntu.com/community/WifiDocs/Adhoc#Encryption Jul 05 10:01:18 Still no WPA Jul 05 10:01:40 But at least it works with the latest wpa_supplicant from OE Jul 05 10:09:49 03Richard Purdie  07master * r32fee2e650 10bitbake.git/lib/bb/ (data_smart.py parse/ast.py): Jul 05 10:09:49 bitbake/ast: Fix ??= vs. ?= handling Jul 05 10:09:49 As the code stands, setting a variable with ??= could result in a ?= Jul 05 10:09:49 variable not overriding it. This patch fixes the issue by allowing Jul 05 10:09:49 the ast to make lookups that ignore any ??= set variables. Jul 05 10:09:49 Signed-off-by: Richard Purdie Jul 05 10:26:59 is it possible to execute a fetch of all packages needed by an image ? Jul 05 10:27:10 I tried bitbake -c fetch helloworld-image without success Jul 05 10:28:12 follow this for the moment for oe-core Jul 05 10:28:14 http://sakrah.homelinux.org/blog/2011/03/using-openembedded-core-to-build-angstrom-for-qemu/ Jul 05 10:28:23 koen's script dont suite me but maybee you Jul 05 10:28:37 woglinde, is that @ me ? Jul 05 10:28:50 is that for me ? Jul 05 10:29:09 yes Jul 05 10:31:55 woglinde: is it possible using 'iw' to enabled encryption, WEP ? Jul 05 10:32:00 woglinde, I'll take a look: I saw that site two days ago, but the script gave me thew idea of non well organized things. Maybe I'm too noob to find the best of this file structure. Jul 05 10:32:26 woglinde, any clue on fetch every source needed for an image to be built ? :$ Jul 05 10:32:31 I only see that you can give key on connect. Jul 05 10:32:32 it organized Jul 05 10:32:35 +is Jul 05 10:33:30 woglinde, I'll see in the afternoon :) Jul 05 10:33:33 thanks a lot. Jul 05 10:33:46 but it dont suites my layouts Jul 05 10:34:03 woglinde, is it good for a beginner ? Jul 05 10:34:04 :) Jul 05 10:34:48 yes Jul 05 10:34:53 it does all you ant Jul 05 10:34:56 but from scratch Jul 05 10:37:00 I used oe with BBFILE_PATTERNS and BBFILE_COLLECTIONS, multiple build directory (once per machine), directory structure as http://wiki.openmoko.org/wiki/OpenEmbedded#BitBake_Overlays (and this seems quite more clear that what I gitcloned yesterday) Jul 05 10:38:48 isn't that from scratch ? Jul 05 10:54:52 woglinde, where to find a "manual" for oe-core ? wiki seems lacking info :( Jul 05 10:55:49 there is no manual yer Jul 05 10:57:23 woglinde, coooooool :) Jul 05 11:08:11 woglinde: what do you dislike more in that a. scripts? the hardcodings maybe? Jul 05 11:08:16 :p Jul 05 11:10:03 ant they layout of layers and where stuff ends up Jul 05 11:10:08 is to static for me Jul 05 11:10:51 I suffered of layers-crosstalk ;) Jul 05 11:11:09 i.e. same recipe was in 3 layers Jul 05 11:11:27 kexec-tools Jul 05 11:11:34 * mwester-laptop likes that term to describe the problem. Jul 05 11:11:56 yes, maybe aliasing would apply too Jul 05 11:12:30 mwester-laptop: we ought to purge meta-nslu2 of kexec-tools :p Jul 05 11:12:39 and more, I'll talk to khem Jul 05 11:12:58 Where are kexec-tools, now? Jul 05 11:13:16 oe-core has 2.02 but without packages splitting Jul 05 11:13:23 and without klibc version Jul 05 11:14:03 I don't think SlugOS uses the klibc version, but I do recall that we wanted the package splitting, no? Jul 05 11:14:30 prolly no need of kdump, just kexec Jul 05 11:14:42 Yep Jul 05 11:14:58 +1 Jul 05 11:20:34 What's our policy on tagging in the OE repo? (I'd like to tag a set of versions for ease of reference.) The alternative would be to create a branch... which seems heavier-weight to me... Jul 05 11:24:03 mwester-laptop: afaik, anybody can create a branch and I can't see any reason why the same shouldn't apply to tagging. Jul 05 11:24:48 * pb_ rebooting, bbiab Jul 05 11:31:32 re Jul 05 11:31:55 that's better, 8G more ram now. hopefully that will be enough for evolution to not swap. Jul 05 11:32:07 :o Jul 05 11:52:15 is there a meta-* list to see if someone wrote a meta-mini2440 ? Jul 05 11:54:53 alx I doubt someone has it done now Jul 05 11:56:10 ok. so let's start with mkdir meta-mini2440 Jul 05 12:00:42 im having a error with mpfr-native configure gmp.h can't be found, or is unusable. any thoughts on how to rectify this? Jul 05 12:02:40 I had similar problems yesterday. how high is your BB_NUMBER_THREADS ? sounds silly, but did you check all software prerequisites ? http://wiki.openembedded.net/index.php/OEandYourDistro Jul 05 12:03:09 Task 319 ? Jul 05 12:03:24 of threads... Jul 05 12:03:25 check your local.conf for BB_NUMBER_THREADS. Jul 05 12:03:38 i think mine is not doing threads Jul 05 12:03:42 mpfr-native depends on gmp-native, so I assume, that the gmp.h should be there Jul 05 12:03:44 oh threads* Jul 05 12:04:29 there are some race. using BB_NUMBER_THREADS="1" I have successfully passed that. Jul 05 12:05:06 i have threads undecalred Jul 05 12:05:21 at the moment Jul 05 12:05:56 try to bitbake gmp-native first Jul 05 12:06:05 then continue the build Jul 05 12:06:06 mmm did you check all software prerequisites ? (sounds silly :P) Jul 05 12:06:21 yeah i installed everything on the wiki Jul 05 12:08:21 follow ynezz : bitbake gmp-native Jul 05 12:08:25 works ? Jul 05 12:09:46 gimmie a few min Jul 05 12:09:59 my builder is teh molases Jul 05 12:11:13 if its built already would it just go by quick? Jul 05 12:12:02 yep Jul 05 12:12:06 "quicker" Jul 05 12:12:09 parsing recipes still :P Jul 05 12:12:28 takes my machine like 5min to parse them Jul 05 12:17:37 hm it apears from declaring bitbake threads to 0 in the local.conf and retrying did it Jul 05 12:17:47 its onto bison already Jul 05 12:18:44 er wait.. task 135 its just staging Jul 05 12:19:01 :P il cancel it out and bitbake gmp-native Jul 05 12:29:01 built fine Jul 05 12:29:14 back to building hello Jul 05 12:29:23 just to be sure, crossscripts/xml2-config should actually have a libdir/incdir pointing into sysroot of my target? - because libxml2-config fails to do so ... Jul 05 12:29:45 03Paul Eggleton  07master * r370fc603d7 10bitbake.git/lib/bb/cooker.py: (log message trimmed) Jul 05 12:29:45 bitbake: track 'overlayed' recipes Jul 05 12:29:45 Recipes that have been 'overlayed' (where there is a recipe in another Jul 05 12:29:45 layer where that layer has a higher priority) are now listed within Jul 05 12:29:45 cooker.overlayedlist for use in bitbake-layers. This is a dict with Jul 05 12:29:46 keys of the topmost (highest priority) recipe file. Jul 05 12:29:47 Signed-off-by: Paul Eggleton Jul 05 12:29:50 03Paul Eggleton  07master * r296c83cc22 10bitbake.git/bin/bitbake-layers: (log message trimmed) Jul 05 12:29:50 bitbake-layers: add command to flatten layers into one Jul 05 12:29:50 Takes the current layer configuration and builds a "flattened" directory Jul 05 12:29:50 containing the contents of all layers, with any overlayed recipes removed Jul 05 12:29:50 and bbappends appended to the corresponding recipes. Note that some manual Jul 05 12:29:50 cleanup may still be necessary afterwards, in particular: Jul 05 12:29:50 * where non-recipe files (such as patches) are overwritten (the flatten Jul 05 12:29:58 03Paul Eggleton  07master * r115b89fa27 10bitbake.git/lib/bb/cooker.py: (log message trimmed) Jul 05 12:29:58 bitbake/cooker: implement layer dependencies, make priority optional Jul 05 12:29:58 Implement (optionally versioned) dependencies between layers, and if layer Jul 05 12:29:59 priorities are not specified using BBFILE_PRIORITY_layername (now Jul 05 12:29:59 optional) then work out the layer priority based on dependencies. Jul 05 12:29:59 Define LAYERDEPENDS_layername in layer.conf to specify the dependencies Jul 05 12:30:00 of a layer (list of layer names, split with spaces in the usual way); Jul 05 12:30:00 03Paul Eggleton  07master * redacf98cce 10bitbake.git/bin/bitbake-layers: Jul 05 12:30:01 bitbake-layers: fix sorting of show_appends output Jul 05 12:30:01 Sort packages alphabetically but ensure appends are left in their Jul 05 12:30:02 original order (layer priority). Jul 05 12:30:02 Signed-off-by: Paul Eggleton Jul 05 12:30:14 mlip doesnt have libxml2 an .pc file Jul 05 12:30:16 use that Jul 05 12:30:25 Signed-off-by: Richard Purdie Jul 05 12:30:25 03Paul Eggleton  07master * rf0c2175dc9 10bitbake.git/bin/bitbake-layers: Jul 05 12:30:25 bitbake-layers: add show_overlayed action Jul 05 12:30:25 Add a show_overlayed action to list overlayed recipes. Jul 05 12:30:26 Signed-off-by: Paul Eggleton Jul 05 12:30:26 Signed-off-by: Richard Purdie Jul 05 12:30:27 instead of -config Jul 05 12:30:53 woglinde, ah that would make sense then, let me check (just tinkering around with an python-lxml recipe) Jul 05 12:33:48 woglinde, libxml-2.0.pc is also screwed for the target ;) Jul 05 12:34:01 man , python extensions cross compiling really sucks Jul 05 12:35:53 woglinde, actually, the targets .pc looks fine, but only if you use it on the target, so it cannot be used for cross-compiling Jul 05 12:41:00 03Dexuan Cui  07master * rf5ba7c795d 10bitbake.git/ (bin/bitbake-layers lib/bb/cooker.py): (log message trimmed) Jul 05 12:41:00 bitbake/cooker, bitbake-layers: show the .bbappend files that matches no existing .bb recipe Jul 05 12:41:00 This patch moves the logic of show_appends_with_no_recipes from bitbake-layers Jul 05 12:41:00 into bitbake. By default, a fatal message is printed; we can also define a variable Jul 05 12:41:00 BB_DANGLINGAPPENDS_WARNONLY to make the message only a warning(the variables Jul 05 12:41:01 could be defined in conf/local.conf with a value "yes", "true" or "1"). Jul 05 12:41:02 Signed-off-by: Dexuan Cui Jul 05 12:46:18 has anyone experienced the following error: Fetch failed: Unable to fetch URL svn://code.ros.org/svn/opencv/branches/2.2;module=opencv;proto=https from any source Jul 05 12:52:54 woglinde, nvm about libxml2, I found the right -config in STAGING_BINDIR_CROSS Jul 05 12:53:59 but I can't find any cross .pc files Jul 05 12:56:33 mlip okay Jul 05 12:57:17 woglinde, do you know whether there exist something like cross-compile .pc files ? Jul 05 12:58:12 Hi all, need some help, thanks in advance for your replies Jul 05 12:58:14 | * resolve_conffiles: Existing conffile /home/guilig/dev/OE/am3517evm-tmp/rootfs/x11-image/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /home/guilig/dev/OE/am3517evm-tmp/rootfs/x11-image/etc/device_table-opkg. Jul 05 12:58:14 | * opkg_install_cmd: Cannot install package locale-base-en-us. Jul 05 12:58:14 NOTE: package x11-image-1.0-r0: task do_rootfs: Failed Jul 05 12:58:14 ERROR: Function 'do_rootfs' failed (see /home/guilig/dev/OE/am3517evm-tmp/work/am3517-evm-angstrom-linux-gnueabi/x11-image-1.0-r0/temp/log.do_rootfs.9298 for further information) Jul 05 12:58:14 ERROR: Task 14 (/home/guilig/dev/OE/openembedded/recipes/images/x11-image.bb, do_rootfs) failed with exit code '1' Jul 05 12:58:14 ERROR: '/home/guilig/dev/OE/openembedded/recipes/images/x11-image.bb' failed Jul 05 13:15:02 nobody can help me ? Jul 05 13:18:36 guilig: Did you disable generating locales in local.conf? Jul 05 13:18:49 yes Jul 05 13:33:33 03Paul Eggleton  07master * r0f45fcab5c 10bitbake.git/bin/bitbake-layers: Jul 05 13:33:33 bitbake-layers: fix error on startup caused by recent cooker change Jul 05 13:33:33 Fix bitbake-layers "TypeError: 'NoneType' object is not iterable" error on Jul 05 13:33:33 startup. Commit f3be8e9a7df13cc11ffc8fc667efaf2db96a7c38 changed to expect Jul 05 13:33:33 prefile and postfile to be populated and no longer expects file. Jul 05 13:33:33 Signed-off-by: Paul Eggleton Jul 05 13:33:34 Signed-off-by: Richard Purdie Jul 05 13:33:43 03Paul Eggleton  07master * r9c2a064ab7 10bitbake.git/bin/bitbake-layers: Jul 05 13:33:43 bitbake-layers: fix sorting by package name for cooker change Jul 05 13:33:43 Structure of cooker was changed since the patch for bitbake commit Jul 05 13:33:43 edacf98cceb2fe1275042595d3fce6822fa411ca was created - cooker.pkg_pn now Jul 05 13:33:43 has string keys, so sort it accordingly. Jul 05 13:33:43 Signed-off-by: Paul Eggleton Jul 05 13:33:44 Signed-off-by: Richard Purdie Jul 05 13:33:50 hi all ... i need some help Jul 05 13:34:07 can you be any more specific? Jul 05 13:35:39 i am unable to fetch URL svn://code.ros.org/svn/opencv/branches/2.2;module=opencv;proto=http Jul 05 13:37:48 that's very sad Jul 05 13:38:36 i am able to go directly to the site and fetch it ... but I can't fetch it via my bitbake recipes Jul 05 13:38:55 well, what happens if you try to check it out manually? Jul 05 13:39:12 i did that ... and I got the same error Jul 05 13:39:37 no, I mean, check it out manually using svn without having bitbake involved at all Jul 05 13:40:07 i have successfully checked it out manually Jul 05 13:40:15 using what command? Jul 05 13:40:24 svn co URL Jul 05 13:40:32 and what URL, specifically? Jul 05 13:41:05 http://code.ros.org/svn/opencv/branches/2.2 Jul 05 13:41:32 well, I just tried that url and it doesn't work for me Jul 05 13:41:58 I just tried fetching the exact svn:// uri posted above with bitbake and it worked for me Jul 05 13:42:19 I get: Jul 05 13:42:20 pb@phil-desktop:~/tmp$ svn co http://code.ros.org/svn/opencv/branches/2.2 Jul 05 13:42:21 svn: Repository moved permanently to 'https://code.ros.org/svn/opencv/branches/2.2'; please relocate Jul 05 13:42:21 pb@phil-desktop:~/tmp$ Jul 05 13:43:26 so, who knows what craziness is going on at ros.org, but it doesn't sound as though bitbake is doing anything wrong Jul 05 13:43:58 03Joshua Lock  07master * r48e64acaae 10bitbake.git/lib/bb/ui/ (8 files in 2 dirs): (log message trimmed) Jul 05 13:43:58 hob: re-designed interaction and implementation Jul 05 13:43:58 Highlights include: Jul 05 13:43:58 * Atempted GNOME HIG compliance Jul 05 13:43:58 * Simplified UI and interaction model Jul 05 13:43:58 * Sorting and type to find in tree views Jul 05 13:43:58 * Preferences dialog to modify local settings Jul 05 13:44:08 the protocol was https in the original URI, the second example cboyd_ posted had http Jul 05 13:44:49 oh, right Jul 05 13:45:02 so, not sure what's going on Jul 05 13:45:11 so you tried with https and it worked. cboyd tried with https using bitbake and it didn't, then he tried with http by hand and it did. Jul 05 13:45:16 cboyd_: are you operating behind some kind of proxy/firewall? Jul 05 13:46:31 thanks regarding the https Jul 05 13:46:36 crumbs, both gnome 3 shell and unity are full of suck Jul 05 13:46:40 * pb_ throws his computer out of the window Jul 05 13:46:50 i am still having trouble Jul 05 13:47:30 pb_: I've just put Ubuntu 11.04 on my laptop. Undecided, verging on unhappy with Unity :( Jul 05 13:47:40 Need to look into a tiling wm, I think. Jul 05 13:47:57 celston: I think I might be able to live with unity on a single screen but it seems not to work very well with multiple monitors Jul 05 13:48:27 for some reason I don't quite understand, it puts the "launcher bar" on the left hand side of my right-hand monitor (i.e. in the middle) and hides it whenever a window goes near Jul 05 13:48:44 so, if I accidentally leave a window straddling the two screens, the launcher disappears which is rather annoying Jul 05 13:48:57 The delay to open the launcher is slooow as well, I'm sure it's configurable. Jul 05 13:49:28 I also don't really like this "maximize the window if it'sdragged to the top of the screen" thing that they both seem to do Jul 05 13:49:47 How do you feel about mac os style menu bars? Jul 05 13:50:06 undecided. I never liked them that much on the Mac, but I never used a Mac intensively enough to really get used to them. Jul 05 13:50:12 I don't mind the maximise behaviour. Jul 05 13:50:19 maybe if I worked with them every day I could get to like them, I dunno Jul 05 13:50:57 Intuitively it should be quicker to always mouse to the same point to access the current window's menu, but I don't think my brain has adjusted yet. Jul 05 13:51:26 yeah. and of course it's further, if you were previously working in a window near the bottom of the screen (or on the other screen) Jul 05 13:51:29 celston: you can go back to a sane window manager fwiw Jul 05 13:51:40 that seems to be another area where the multimonitor support is a bit lacking Jul 05 13:51:52 RP__: I suspect I will. Jul 05 13:53:12 celston: http://dwm.suckless.org/ ftw ;-) Jul 05 13:55:12 thhp: "Because dwm is customized through editing its source code, it’s pointless to make binary packages of it. This keeps its userbase small and elitist. No novices asking stupid questions. " chortle Jul 05 13:55:55 heh Jul 05 13:56:15 written in C though, how lame. surely it should be in assembler. Jul 05 13:59:19 heh Jul 05 14:01:45 hi mickey Jul 05 14:01:51 hey pb_ Jul 05 14:11:48 hi mickeyl Jul 05 14:12:21 moinmoin woglinde Jul 05 14:30:10 lul im stuck at cross-initial for hours Jul 05 14:30:13 my computer sucks Jul 05 14:30:39 was thumbing through craigslist for cheap computer to buy :D Jul 05 14:33:32 comes with windows and malware scanner! :P just devalued it in my opinion heh Jul 05 15:12:04 03Yiping Wang  07master * r400736229d 10openembedded.git/recipes/ethtool/ (ethtool.inc ethtool_2.6.35.bb ethtool_2.6.36.bb): Jul 05 15:12:04 ethtool: Add 2.6.26 Jul 05 15:12:04 Upgrade to 2.6.36 verion for ipsec 2c2pf support. Jul 05 15:12:04 Signed-off-by: Yiping Wang Jul 05 15:12:04 Signed-off-by: Tom Rini Jul 05 15:35:57 03Richard Purdie  07master * rc8f46dfcc2 10bitbake.git/lib/bb/ui/depexp.py: Jul 05 15:35:57 ui/depexp: If we're parsing zero files we need to ensure the cache progress bar gets hidden Jul 05 15:35:57 Signed-off-by: Richard Purdie Jul 05 15:45:22 hmmm udev doesn't seem to be very core Jul 05 15:45:49 looking at the commits shows that some people pushed without 2 acks Jul 05 15:45:54 => I'll do the same Jul 05 15:47:16 + a lot of people don't use udev Jul 05 15:47:20 at least it seems Jul 05 15:59:04 I think angstrom uses udev. I don't know if any other distros do. Jul 05 16:02:09 SHR doesn't use udev Jul 05 16:02:18 so can I still push Jul 05 16:02:22 it's tested etc... Jul 05 16:05:00 basically I've 2 changes for udev: Jul 05 16:05:13 *one to change the maximum kernel from 2.6.26 to 2.6.27 Jul 05 16:05:17 for making bug work Jul 05 16:05:22 bug is a machine Jul 05 16:05:30 the other is rather minimal Jul 05 16:05:48 it's RRECOMMANDS_${PN}_bug on their udev rules Jul 05 16:07:20 maximum kernel is for udev compat Jul 05 16:07:22 there was Jul 05 16:07:32 if kenrel -lt 2.6.27 Jul 05 16:07:36 I changed to 2.6.28 Jul 05 16:07:47 I verified that on target Jul 05 16:07:56 with the linux-bug 2.6.27 kernel Jul 05 16:08:08 I think I'll push Jul 05 16:10:10 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r2b7ffeb0ff 10openembedded.git/recipes/udev/ (8 files in 2 dirs): Jul 05 16:10:10 udev-compat: switch the maximum kenrel version from 2.6.26 to 2.6.27 Jul 05 16:10:10 Without that patch the bug machine,which has a 2.6.27 hangs at boot here: Jul 05 16:10:10 Checking for built-in Bluetooth: no Jul 05 16:10:10 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 16:10:16 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r301be5cccd 10openembedded.git/recipes/tor/ (files/tor.init tor_0.1.1.26.bb): Jul 05 16:10:17 tor: quiet the init script(else it prints to the console) Jul 05 16:10:17 Without that patch the tor init script print to the console, Jul 05 16:10:17 even when the init script has finished to run. Jul 05 16:10:17 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 16:10:17 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r99586e7205 10openembedded.git/recipes/udev-rules-buglabs/ (bug-udev/10-mx31.rules bug-udev/30-BUG.rules bug-udev_0.1.bb): Jul 05 16:10:19 bug-udev: add support for bug1 Jul 05 16:10:20 10-mx31.rules and 30-BUG.rules were imported from the udev recipes Jul 05 16:10:20 directory( recipes/udev/udev/bug/ ) Jul 05 16:10:20 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 16:10:20 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r8be60a2d5a 10openembedded.git/recipes/tor/tor_0.2.1.30.bb: Jul 05 16:10:21 tor: add 0.2.1.30 version Jul 05 16:10:21 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 16:10:22 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r4852371bf8 10openembedded.git/recipes/prboom/ (prboom-devconfig.bb prboom-devconfig/nokia900/prboom.cfg): (log message trimmed) Jul 05 16:10:22 prboom-devconfig: add nokia900 config Jul 05 16:10:23 Note that the prboom doesn't look for the system configuration if Jul 05 16:10:23 you don't tell it to do so by running it the following way: Jul 05 16:20:00 I've to go Jul 05 16:38:29 03Eric BENARD  07master * r5336b30b9f 10openembedded.git/conf/distro/include/angstrom-2010-preferred-versions.inc: Jul 05 16:38:30 angstrom-2010: prefer busybox 1.18.5 Jul 05 16:38:30 Signed-off-by: Eric Bénard Jul 05 16:38:30 Cc: Koen Kooi Jul 05 16:38:34 03Eric BENARD  07master * rff9ad5e1a1 10openembedded.git/conf/distro/include/preferred-slugos-versions.inc: Jul 05 16:38:35 slugos: prefer busybox 1.18.5 Jul 05 16:38:35 Signed-off-by: Eric Bénard Jul 05 16:38:35 Cc: Mike Westerhof Jul 05 16:38:35 Acked-by: Mike Westerhof Jul 05 16:38:44 03Eric BENARD  07master * r9fb794af4a 10openembedded.git/recipes/busybox/ (11 files in 3 dirs): Jul 05 16:38:45 busybox-1.18.4: remove old version Jul 05 16:38:45 Signed-off-by: Eric Bénard Jul 05 16:38:45 03Eric BENARD  07master * re0541dc6fd 10openembedded.git/recipes/busybox/ (11 files in 3 dirs): Jul 05 16:38:45 busybox: add 1.18.5 Jul 05 16:38:45 which is the latest stable version (1.18.4 + fixes) Jul 05 16:38:46 Signed-off-by: Eric Bénard Jul 05 18:38:32 awe desktop-file-utils-0.16/configure: line 11285: syntax error near unexpected token `DESKTOP_FILE_UTILS,' Jul 05 18:38:44 whys it compiling that for helloworld anyway? Jul 05 19:01:38 nullbody because of kernel bluez maybee Jul 05 19:01:41 hi risca Jul 05 19:03:42 hi woglinde =) Jul 05 19:05:14 * risca wishes he could do more oe work instead of working, fixing car, fixing freezer, fixing bad toe, etc. ^^ Jul 05 19:05:30 risca hm Jul 05 19:21:05 googling said DESKTOP_FILE_UTILS is glib so i guess i'll try to bitbake that? Jul 05 19:21:46 could have swore i saw eglib compile not long ago Jul 05 20:01:48 woglinde, hi Jul 05 20:07:29 hmmm Jul 05 20:07:31 http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/udev/udev_171.bb Jul 05 20:07:39 are the _bug ok? Jul 05 20:07:44 pb_ ping? Jul 05 20:07:58 should I set it to ARCH_bug = ${MACHINE_ARCH} ? Jul 05 20:08:35 because I blindly followed the comments(for both udev compat and the RRECOMMANDS_${PN} Jul 05 20:19:00 anyone? Jul 05 20:23:03 Architecture: bug Jul 05 20:23:05 seem ok Jul 05 20:23:19 ah it's 168 Jul 05 20:23:21 hmmm Jul 05 20:23:23 let me look Jul 05 21:06:26 * GNUtoo doesn't understand a thing, I bitbaken udev 171 and I've 168 on target rootfs Jul 05 21:07:16 what exactly do you mean by 'bitbake udev 171'? Jul 05 21:07:40 bitbake udev-171? that's not going to do what you want. it could end up building both versions in a single build and things will get very confused Jul 05 21:07:57 bitbake udev Jul 05 21:08:03 and seeing that: Jul 05 21:08:16 NOTE: package udev-171-r1: task do_package: Started Jul 05 21:08:20 and such messages Jul 05 21:08:25 with only udev-171 Jul 05 21:08:49 maybe it's doing that: Jul 05 21:08:55 I bitbaken udev 168 before Jul 05 21:08:57 it takes that Jul 05 21:09:05 because it's machine arch Jul 05 21:09:13 thats odd, considering the creation of hte package feed should move the old version into the attic Jul 05 21:09:15 and 171 isn't? Jul 05 21:09:19 ah, that could be Jul 05 21:09:53 btw I'm not sure I made a mistake or not: Jul 05 21:09:54 http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/udev/udev_171.bb Jul 05 21:10:01 I added the _bug stuff Jul 05 21:10:08 following blindly the comments Jul 05 21:10:13 is it an issue Jul 05 21:10:19 should it become MACHINE_ARCH? Jul 05 21:10:21 for bug Jul 05 21:10:33 like PACKAGE_ARCH_bug = ${MACHINE_ARCH} Jul 05 21:10:46 with the proper quotes Jul 05 21:11:23 ineded I've a machine arch udev 168 Jul 05 21:11:31 interesting Jul 05 21:11:56 yes the older udevs had support for some rules specific to my machine Jul 05 21:12:03 so it became machine arch Jul 05 21:13:52 kergoth, what should I do now? Jul 05 21:14:01 add MACHINE_ARCH for my machine? Jul 05 21:14:36 hmmm Jul 05 21:14:43 there is 168 also in armv6 Jul 05 21:14:45 strange Jul 05 21:15:14 udev-dbg_168-r2.9_armv6.ipk Jul 05 21:15:15 and: Jul 05 21:15:20 udev-dev_168-r2.9_armv6.ipk Jul 05 21:15:28 and in bug there is: Jul 05 21:15:51 I'll pastebin Jul 05 21:15:58 dunno, never had to deal with the difference between arch and machine specific packages wrt opkg comparisons and versioning, sounds unpleasant Jul 05 21:16:10 http://pastie.org/private/dy1m3f8bocz47pwr2veg Jul 05 21:18:17 kergoth, so I add udev-171 as machine arch to fix at least the _bug ? Jul 05 21:18:30 i'd guess so, but don't really know Jul 05 21:19:43 thanks a lot Jul 05 21:19:56 unfortunately no ones seem to know well udev.... Jul 05 21:20:15 hmmm Jul 05 21:20:33 I asked for ages for the udev-compat thing Jul 05 21:20:41 I spent days trying Jul 05 21:20:53 GNUtoo: whats the problem Jul 05 21:20:55 and at the end I decided to bruteforce the variables Jul 05 21:21:12 hi khem Jul 05 21:21:14 2 issues: Jul 05 21:21:37 1) http://cgit.openembedded.org/cgit.cgi/openembedded/tree/recipes/udev/udev_171.bb has _bug and no PACKAGE_ARCH_bug = "bug" Jul 05 21:21:39 should I add it Jul 05 21:21:41 ? Jul 05 21:21:52 2) bitbake udev shows udev 171 is bitbaken Jul 05 21:22:05 but because udev168 is machine arch for my machine and was bitbaken before Jul 05 21:22:17 udev 168 ends in my image Jul 05 21:22:48 for the udev-compat I know know how it works Jul 05 21:23:19 PACKAGE_ARCH_bug = "bug" why ? Jul 05 21:23:27 because there is _bug stuff Jul 05 21:23:44 I blindly followed the advises in the comments Jul 05 21:23:51 and added the _bug overrides Jul 05 21:24:38 there are some bug specific changes in recipes which is covered with bug override Jul 05 21:24:39 else it would mess up multi-machine builds Jul 05 21:24:40 which is fine Jul 05 21:24:41 Sorry if this is out of the blue but I get errors like the following in my OE build: Jul 05 21:24:45 NOTE: The localpath does not exist 'sources/downloads/sqlite-2.8.17.tar.gz' Jul 05 21:25:00 Have I messed up my tmp directory somehow? Jul 05 21:25:21 joelagnel: is it oe-core or classic oe Jul 05 21:25:29 khem: classic oe Jul 05 21:25:42 khem, ah but let's say you build with machine bug, and then another armv6 machine, you get the _bug at first build and at the second build, it doesn't rebuild and you get _bug too Jul 05 21:25:53 cd into dnld dir and see if you have the tar and checksum file Jul 05 21:26:16 I deleted the tar because it was throwing check sum errors, hoping it would fetch it again Jul 05 21:26:20 I did a -c clean too Jul 05 21:26:37 joelagnel, try to find the .md5 that is in the same dir than the tarball Jul 05 21:26:37 GNUtoo: I see the point so as long as you do not build for say two different armv6 machines one of which is bug its ok Jul 05 21:26:57 khem: hey, wb Jul 05 21:26:59 joelagnel: delete sqlite-2.8.17.tar* Jul 05 21:27:00 khem, but angstrom build for tons of machine and uses udev Jul 05 21:27:08 hi ant__ Jul 05 21:27:08 GNUtoo: yes true Jul 05 21:27:17 so I add the MACHINE_ARCH to fix Jul 05 21:27:22 but only for bug Jul 05 21:27:26 not for all Jul 05 21:27:30 GNUtoo: those bug override essentially makes it machine specific Jul 05 21:27:32 else they would build 1 udev per device Jul 05 21:27:35 yes Jul 05 21:27:36 indeed Jul 05 21:27:39 so I add it Jul 05 21:27:59 the right thing to ask would be does bug need to be different ? Jul 05 21:28:06 it does Jul 05 21:28:12 xorg.conf expect some simlinks Jul 05 21:28:18 and the java stuff that will come too Jul 05 21:28:32 and also most important Jul 05 21:28:36 it need the compat udev Jul 05 21:28:44 because it has a 2.6.27 kernel Jul 05 21:28:55 PACKAGE_ARCH = "${MACHINE_ARCH}" Jul 05 21:28:58 else the machine doesn't boot Jul 05 21:29:02 no rather that: Jul 05 21:29:20 PACKAGE_ARCH_bug = "bug" Jul 05 21:29:29 else it will make it machine arch for each machine Jul 05 21:29:32 which is very bad Jul 05 21:29:34 ah it builds now, sweet. Jul 05 21:29:39 angstrom people would get furious Jul 05 21:29:53 What could have gone wrong (just so I'm careful next time)? Jul 05 21:29:55 because their build time would increase a lot Jul 05 21:30:06 joelagnel, the download could have been corrupted Jul 05 21:30:33 GNUtoo: oh ok, I'm getting on a whole bunch of recipes actually Jul 05 21:30:35 joelagnel: wild wild web can give you crap sometimes and it might have corrupted the tar file Jul 05 21:30:49 joelagnel: if you use oe-core then you could do bitbake -ccleanall Jul 05 21:30:55 Should I just cleanup the download directory? Jul 05 21:30:57 Jul 05 21:31:03 not everything Jul 05 21:31:13 rm source/downloads/* ? Jul 05 21:31:13 do it selectively Jul 05 21:31:24 well if you have time and bandwidth go for it Jul 05 21:31:36 but you do not need swords to kill flies Jul 05 21:31:46 khem: would you mind to remove kexec-tools and klibc stuff from meta-nslu2 ? Jul 05 21:32:07 kexec-tools is in oe-core (though without grained packaging) Jul 05 21:32:10 ant__: I would be happy to do that is it available on a different layer ? Jul 05 21:32:15 ah oe-core ? Jul 05 21:32:17 khem: ah ok Jul 05 21:32:21 klibc is coming... Jul 05 21:32:32 just need your help for packaging :) Jul 05 21:32:36 thank you so much for your help! Appreciate it. Jul 05 21:32:47 ant__: so once it makes into oe-core with all changes that are in slugos Jul 05 21:32:48 I brutally install klcc-cross by hand Jul 05 21:32:51 then I would drop it Jul 05 21:32:58 is GitHub the primary repository for oe-core? I'm not seeing an oe-core branch on the OE git server. Jul 05 21:33:01 khem: ok Jul 05 21:33:21 zenlinux: hmmm github is preferred way for pulling Jul 05 21:33:32 but it does not mean oe-core is hidden on oe.org Jul 05 21:33:33 khem, ok, just verifying Jul 05 21:33:52 http://git.openembedded.org/cgit.cgi/openembedded-core/ Jul 05 21:34:01 khem, ah, thanks Jul 05 21:34:12 its not a branch its a entirely separate repo Jul 05 21:34:19 gotcha Jul 05 21:34:33 GNUtoo: so whats the second problem Jul 05 21:35:11 a problem fixed by it beeing MACHINE_ARCH Jul 05 21:35:32 MACHINE_ARCH_bug fix it for both Jul 05 21:35:38 *fixes both issues Jul 05 21:35:47 khem: once you've time: http://paste.debian.net/121998/ Jul 05 21:36:30 GNUtoo: RDEPENDS_udev_bug += "udev-compat" Jul 05 21:36:30 do_unpack_append_bug() { bb.build.exec_func('do_apply_compat_wrapper', d) Jul 05 21:36:30 } Jul 05 21:36:46 that code is not how it should be done Jul 05 21:36:52 ah? Jul 05 21:36:58 how should it be done then? Jul 05 21:37:06 if bug needs some different udev then let there be a recipe for udev-bug Jul 05 21:37:13 it was done that for : Jul 05 21:37:16 *palmpre Jul 05 21:37:18 *zaurus Jul 05 21:37:21 and then use that istead of classic udev Jul 05 21:37:34 ant__, hi Jul 05 21:37:42 I'm not sure Jul 05 21:37:51 I think the way I did it is correct Jul 05 21:37:58 maybe ant__ can comment? Jul 05 21:37:58 if there are more than 1 machine which needs such special stuff then clearly the recipe should be marked as machine specific Jul 05 21:38:02 GNUtoo: I know the battle:) soo happy we are on 2.6.39 / 3.0 :] Jul 05 21:38:11 lol ok Jul 05 21:38:17 no matter how long it takes to compile since it will be the correct thing Jul 05 21:38:20 ant__, touched the udev recipe often Jul 05 21:38:49 khem, what's the difference? Jul 05 21:38:56 both are machine arch Jul 05 21:39:12 and every kernel < 2.6.38 wants udev-compat Jul 05 21:41:01 khem: in that way I put the arm-...klcc in the right place, where gcc-cross is Jul 05 21:41:11 so I can compile Jul 05 21:41:22 but klcc-cross is unpackaged :/ Jul 05 21:42:00 ant__: is it for oe-core ? Jul 05 21:42:09 yes Jul 05 21:42:13 udev 171 fails Jul 05 21:42:20 I'll try harder to make it work Jul 05 21:42:26 ant__: its better to be in native_sysroot/usr/bin imo Jul 05 21:42:57 udev-compat yes is needed for older kernels Jul 05 21:44:20 khem: it ends in andrea@mizar /oe/setup-scripts/build/tmp-angstrom_2010_x-eglibc/sysroots/i686-li Jul 05 21:44:20 nux/usr/bin/armv5te-angstrom-linux-gnueabi $ ./arm-angstrom-linux-gnueabi-klcc Jul 05 21:45:04 ant__: it will probably work but that is toolchain staging area Jul 05 21:45:22 I am not sure if bitbake -cclean gcc-cross will nuke everything under it Jul 05 21:45:39 nope Jul 05 21:45:56 you can put it in usr/bin/ Jul 05 21:46:11 instead of usr/bin/armv5te-angstrom-linux-gnueabi Jul 05 21:46:26 which might not work Jul 05 21:46:42 does it invoke gcc internally Jul 05 21:46:45 hm.. isn't /usr/bin for klcc ? Jul 05 21:47:00 this ought to be klcc-cross Jul 05 21:47:09 I meant /usr/bin Jul 05 21:47:21 you can install libklibc-dev on host in theory Jul 05 21:47:28 toolchain is in /usr/bin/armv5te-angstrom-linux-gnueabi Jul 05 21:48:09 FILESPATH =. "${FILE_DIRNAME}/klibc-${PV}:" Jul 05 21:48:18 that I think wont work Jul 05 21:48:28 well, last sysroot change took earth under my feet :p Jul 05 21:48:49 the armv5- sysroot has disappeared Jul 05 21:49:17 I have only i686-linux and poodle (and poodle-tcbootstrap fwiw) Jul 05 21:49:18 it might work Jul 05 21:49:31 yes Jul 05 21:49:37 i686-linux = native sysroot Jul 05 21:49:46 poodle = target sysroot Jul 05 21:49:57 poodle-tcbootstrap = toolchain intermediate sysroot Jul 05 21:50:10 of interest to u are i686-linux and poodle Jul 05 21:51:40 khem: the recipes are building fine fwiw (was FILESPATH) Jul 05 21:52:37 ant__: yeah thats ok Jul 05 21:53:33 ant__, how can I make udev 171 work with my device? Jul 05 21:53:47 I backported sys-accept4 to 2.6.27 Jul 05 21:54:06 sorry, no idea of the new requirements Jul 05 21:54:08 I made 2.6.27 use the compat udev Jul 05 21:54:12 and it still fails Jul 05 21:54:43 udev 168 works fine with compat Jul 05 21:54:46 but not 171 Jul 05 21:56:37 according to http://git.kernel.org/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=README I need quite some stuff Jul 05 21:56:40 let me look Jul 05 21:58:51 GNUtoo: with that old kernel you need lot more Jul 05 21:59:00 GNUtoo: you need epoll Jul 05 21:59:08 and signalfd too Jul 05 21:59:11 ohhh no Jul 05 21:59:25 udev goes hand in hand with kernel Jul 05 21:59:27 what should I do Jul 05 21:59:30 ? Jul 05 21:59:33 I need udev rules Jul 05 21:59:39 and sometimes needs support from libc too Jul 05 21:59:41 udev 168 works fine Jul 05 21:59:48 keep using 168 Jul 05 21:59:51 how? Jul 05 22:00:01 you said it work Jul 05 22:00:16 huh Jul 05 22:00:35 khem: now seems in better shape..let see if I can build Jul 05 22:00:44 yes but how do I select 168? Jul 05 22:00:59 pin it Jul 05 22:01:01 I selected it in local.conf Jul 05 22:01:05 but I want it in oe Jul 05 22:01:10 how do I pin it in oe? Jul 05 22:01:16 I add that to machine config: Jul 05 22:01:19 which machine/distro Jul 05 22:01:27 ok I'll try with machine Jul 05 22:01:49 argh.. compiler not found Jul 05 22:02:14 khem, angstrom, bug Jul 05 22:02:44 under oe.dev Jul 05 22:02:45 you could do PREFERRED_VERSION_udev_bug = "168" in include/angstrom-2008-preferred-versions.inc Jul 05 22:02:54 ok Jul 05 22:03:19 who should I ping for the ack of that? Jul 05 22:03:24 koen Jul 05 22:03:28 ouch Jul 05 22:03:30 ok Jul 05 22:03:52 explain it well Jul 05 22:03:56 ok Jul 05 22:04:12 personally I've no idea why it blocks there Jul 05 22:04:16 at bluetooth Jul 05 22:04:22 the only thing I know Jul 05 22:04:28 is that udev fails to work Jul 05 22:04:37 and the dev nodes are not created Jul 05 22:04:39 etc... Jul 05 22:06:05 so I revert my udev change Jul 05 22:06:11 and will send a patch for the pinning Jul 05 22:06:25 and will add support for my machine in udev 168 Jul 05 22:07:09 yes Jul 05 22:07:19 is it possible for you to use newer kernel ? Jul 05 22:07:27 like 2.6.39 or so Jul 05 22:07:36 then 171 will work Jul 05 22:07:43 or rather work better Jul 05 22:07:50 khem, can you edit wiki? Jul 05 22:08:27 khem, I must do the port, I tried but the 2 maintainers are silent Jul 05 22:08:37 I keep pinging like ever 2 weeks or more Jul 05 22:13:28 wb Crofton Jul 05 22:14:52 thanks Jul 05 22:17:52 ka6sox: havent tried in long itme Jul 05 22:19:54 khem, could you some time? Jul 05 22:20:02 ka6sox: sure Jul 05 22:20:04 I have reports of "issues" Jul 05 22:20:16 ka6sox: any particular thing I should try Jul 05 22:20:33 yes, something on the getting started page Jul 05 22:20:47 its a protected page you should be alble to edit Jul 05 22:26:09 ka6sox: Seems not Jul 05 22:26:20 ka6sox: there is no edit knob seen at all Jul 05 22:26:47 you are signed in? Jul 05 22:28:43 ka6sox: yes as a matter of fact I can not edit any page Jul 05 22:33:19 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r97f76f3a67 10openembedded.git/recipes/xinput-calibrator/ (2 files in 2 dirs): Jul 05 22:33:19 xinput-calibrator's pointercal: add support for bug1 Jul 05 22:33:19 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 22:33:24 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * rc653e2f24a 10openembedded.git/recipes/xorg-xserver/ (xserver-xorg-conf/bug/xorg.conf xserver-xorg-conf_0.1.bb): Jul 05 22:33:24 xserver-xorg-conf: switch bug to evdev Jul 05 22:33:24 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 22:33:25 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * ra36350f7e2 10openembedded.git/ (conf/machine/bug.conf recipes/tasks/task-x11.bb): Jul 05 22:33:26 bug: remove the dependency on xf86-video-tslib Jul 05 22:33:26 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 22:33:26 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * ra1b8d5b53c 10openembedded.git/recipes/udev/udev_171.bb: (log message trimmed) Jul 05 22:33:26 udev 171: partially revert commit d5840f76a5c4d145b1a81de48fdd7cb0dc9f2ee ( bug: unbreak udev 171 for bug. ) Jul 05 22:33:26 This commit didn't work: Jul 05 22:33:26 I tested on target but because udev 168 was built before and is machine arch Jul 05 22:33:27 for the bug machine it got selected. Jul 05 22:33:27 Instead selecting udev 171 by making it machine arch for bug made the bug fail Jul 05 22:33:35 to boot Jul 05 22:33:35 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * rd21149de5c 10openembedded.git/recipes/udev/udev_168.bb: Jul 05 22:33:35 udev 168: add support for the bug machine (with compat udev) Jul 05 22:33:35 Signed-off-by: Denis 'GNUtoo' Carikli Jul 05 22:35:03 khem, you aren't logged in Jul 05 22:38:32 ka6sox: yes I am :) Jul 05 22:38:43 orly...interesting. Jul 05 22:39:43 try relogging in... Jul 05 22:39:48 the server doesn't think you are. Jul 05 22:42:14 ka6sox: see http://uclibc.org/~kraj/Screenshot.png Jul 05 22:43:20 ka6sox: relogin does not help Jul 05 22:43:32 okay this is wierd. Jul 05 22:44:20 try now please? Jul 05 22:44:26 I made one small change. Jul 05 22:48:11 ok now it seems to work Jul 05 22:48:29 khem, thanks...now I know whats up...but not why. Jul 05 22:48:41 I'll dig into this more. Jul 05 22:48:53 ok 33% Jul 05 22:49:07 what + why + fix :) Jul 05 22:49:23 I have the "fix" so 66% Jul 05 22:49:32 only 33% unknown. Jul 05 22:49:45 it might be a wrong fix once you know "why" Jul 05 22:50:23 yes, I hope this fix will work...because if it does, will make my job easier. Jul 05 22:52:29 khem: well, I give up... I'll send out a crappy but working recipe instead of a polished non-working one Jul 05 22:52:40 bluelightning: ping Jul 05 22:52:45 it wont go in Jul 05 22:52:52 unless its in best shape Jul 05 22:53:07 khem: another outstanding issue I found: we have two kernel.bbclass atm Jul 05 22:53:13 yes Jul 05 22:53:15 the one in oe-core ignores initramfs Jul 05 22:53:19 if you use meta-oe Jul 05 22:53:19 :/ Jul 05 22:53:37 there is a reason why there is one in meta-oe Jul 05 22:53:44 really? Jul 05 22:53:58 is confusing at least Jul 05 22:54:54 yes something in bitbake to dump the files it read was discussed Jul 05 22:55:04 that can give some idea of what comes from where Jul 05 22:55:19 but frankly same problem can happen with multiple recipes Jul 05 22:55:34 you dont need layers to mess it up Jul 05 22:58:14 khem: ifI do as you say, klcc-cross ends up as a target package Jul 05 22:58:30 even worse Jul 05 22:58:36 like the binaries in initramfs Jul 05 22:58:41 hmm Jul 05 22:58:43 won't work Jul 05 22:58:59 build klcc-cross on the lines of gcc-cross Jul 05 22:59:00 it's a perl wrapper Jul 05 22:59:24 klcc-cross is perl wrapper around gcc-cross right ? Jul 05 22:59:31 yes Jul 05 23:00:17 CC = "\/oe\/setup\-scripts\/build\/tmp\-angstrom_2010_x\-eglibc\/sysroots\/i686\-linux\/usr\/bin\/armv5te\-angstrom\-linux\-gnueabi\/arm\-angstrom\-linux\-gnueabi\-gcc"; Jul 05 23:00:30 yeah Jul 05 23:00:40 so make it inherit cross Jul 05 23:00:45 heh Jul 05 23:00:54 now in oe-core that is impossiuble Jul 05 23:01:12 during builds it can't find the headers Jul 05 23:01:36 export KLCC_INST=${STAGING_DIR_TARGET}/lib/klibc Jul 05 23:01:48 ^^ maybe this is wrong Jul 05 23:02:15 trust me, this is *the* pandora recipe :) Jul 05 23:03:20 which headers Jul 05 23:03:52 moment, you need to see klibc for oe-core Jul 05 23:04:58 http://paste.debian.net/122018/ Jul 05 23:05:42 http://paste.debian.net/122020/ Jul 05 23:10:22 khem: inheriting cross: http://paste.debian.net/122022/ Jul 05 23:11:26 sysroots/i686-linux/kernel Jul 05 23:12:18 which isn't there Jul 05 23:41:26 which mailing list should get patches for meta-openembedded? Jul 06 00:01:18 sakoman_: oe-devel Jul 06 00:01:33 sakoman_: oe-core patches go to oe-core Jul 06 00:25:39 My recipe pulls from a git tree, but when I set SRCREV to anything other than AUTOREV (such as a meaningful revision number), it seems to be pulling from an http link (from angstrom-distribution.org) instead of the git tree Jul 06 00:25:51 Can anyone explain this behaviour? Jul 06 00:26:22 joelagnel: u have mirrors set Jul 06 00:26:28 are you using angstrom ? Jul 06 00:26:33 as distro Jul 06 00:26:37 khem: yes Jul 06 00:26:42 ok Jul 06 00:27:00 so angstrom first groks its own mirrors before going into rest of www Jul 06 00:27:12 Is there anyway to force the recipe to ignore the mirror? Jul 06 00:27:21 yes there is Jul 06 00:27:28 I dont remember OTOH Jul 06 00:27:32 grok mls Jul 06 00:27:44 sure Jul 06 00:27:44 khem: I cc'ed you in case you'd take care of this little puppy ;) Jul 06 00:28:19 Is pulling from HEAD considered bad practice if I can make sure that every commit results in a stable tree? Jul 06 00:28:48 i.e. if SRCREV = ${AUTOREV} Jul 06 00:29:46 usually its ok depends on recipe to recipe Jul 06 00:29:51 good night Jul 06 00:29:58 ant__: gn Jul 06 00:30:19 khem: ty Jul 06 00:51:22 khem, fixed issue...permissions issue that has been fixed. **** ENDING LOGGING AT Wed Jul 06 02:59:57 2011