**** BEGIN LOGGING AT Sun Mar 21 02:59:57 2010 Mar 21 03:04:42 03Holger Hans Peter Freyther  07org.openembedded.dev * r30c7c2f4a6 10openembedded.git/recipes/php/ (4 files in 3 dirs): Mar 21 03:04:43 php: 5.2.13 and 5.3.2 both have flaws in the handling of xmlrpc Mar 21 03:04:43 This is addressing CVE-2010-0397.patch. Mar 21 03:17:04 03Mike Westerhof  07org.openembedded.dev * r3452deb0d9 10openembedded.git/recipes/apex/ (apex-nslu2-16mb_1.5.14.bb apex-nslu2_1.5.14.bb): Mar 21 03:17:05 apex-nslu2: replace endianness tests to not require CONFIG_SITE. Mar 21 03:17:05 Required since CONFIG_SITE is no longer automatically defined Mar 21 03:17:05 by siteinfo.bbclass (now part of autotools.bbclass). Mar 21 03:17:15 03Mike Westerhof  07org.openembedded.dev * r4dd15c895a 10openembedded.git/conf/distro/include/preferred-slugos-versions.inc: Mar 21 03:17:15 preferred-slugos-versions.inc: bump up autotools-native, and lock down autotools Mar 21 03:17:15 This will avoid build problems when newer autotools become available. Mar 21 05:52:14 anyone around? Mar 21 08:59:09 did someone notice lately, that some packages are rebuilt over and over again? This happen ie after EFL bump, but there were quite big .bbclass changes Mar 21 08:59:17 so I'm not sure where to look first Mar 21 09:53:18 hello :) Mar 21 09:53:43 I'm new to openembedded.. do you have a suggestion how to start developing for qt-embedded with it? Mar 21 11:45:36 03Roman I Khimov  07org.openembedded.dev * r7bf5690c6b 10openembedded.git/classes/cmake.bbclass: (log message trimmed) Mar 21 11:45:36 cmake.bbclass: force CMAKE_INSTALL_SO_NO_EXE to be 0 Mar 21 11:45:36 By default cmake guesses whether install shared libaries with exec Mar 21 11:45:36 permissions or not based on host distribution, so this behaviour Mar 21 11:45:36 changes from host to host. OpenEmbedded assumes shared libraries to Mar 21 11:45:36 be with exec permissions (reverse way brakes stripping, for example), Mar 21 11:45:37 so let's force this in cmake configure invocation. Mar 21 11:50:35 03Roman I Khimov  07org.openembedded.dev * r7f2f7b3ad5 10openembedded.git/recipes/openldap/ (openldap-2.4.17/openldap-m4-pthread.patch openldap_2.4.17.bb): Mar 21 11:50:35 openldap: update version 2.4.17 to 2.4.21 Mar 21 11:50:35 Signed-off-by: Roman I Khimov Mar 21 13:30:49 Hi Guys - I'm struggling with OE (x11-image, angstrom) built on an iPAQ h5550. Device boots, I can see the kernel messages on the serial port, angstrom splash on the screen, but X never starts, neither does the getty on the serial port. Mar 21 13:31:17 The getty seems to be set up right in the inittab. Mar 21 13:33:16 Disconnecting and reconnecting the USB cable causes further gadget ethernet log entries on the serial port - so the device doesn't seem to be dead. Mar 21 13:37:27 sorry, whats getty? Mar 21 13:39:52 aditya_1212, http://en.wikipedia.org/wiki/Getty_%28Unix%29 Mar 21 13:40:16 Runs on the serial port, should provide a serial terminal Mar 21 13:42:19 Hm, though I've just noticed that I get the kernel space boot messages, but none of the userspace ones... Mar 21 13:44:37 oh Mar 21 13:44:47 so you dont get the shell on the serial port Mar 21 13:44:56 nope. Mar 21 13:45:00 hmm Mar 21 13:45:24 this is just a vanilla image oyu just made? Mar 21 13:45:44 yeah Mar 21 13:46:22 Looks like the getty is provided by tinylogin Mar 21 13:46:50 I dont have my bboard near me, cant check :/ Mar 21 13:47:06 but I can't imagine the image for iPaQ uses something special Mar 21 13:47:37 * MWelchUK hates the fact that the iPAQ needs to be disconnected from the cradle to get to the reset button! Mar 21 13:48:29 lol Mar 21 13:48:37 near the dock connector or something? Mar 21 13:48:58 On the bottom edge. Mar 21 13:49:04 ah Mar 21 13:49:33 should just drill a tiny hole into your dock and then your disk Mar 21 13:49:37 desk* Mar 21 13:49:40 and make sure they are aligned Mar 21 13:49:45 easy peasy :P Mar 21 13:51:43 Not that easy - the dock has a mechanism built into it to allow a sleeved or unsleeved iPAQ to be placed in it - it isn't hollow. Mar 21 13:51:59 I was mostly joking Mar 21 13:52:04 that would be a horrible thing to try lol Mar 21 13:57:20 cant imagine why your serial terminal isn't showing up though Mar 21 13:57:27 gm Mar 21 13:57:30 is it hooked up to the local network? Mar 21 13:57:33 hey likewise Mar 21 13:57:43 I usually ssh into my bboard for most of my work Mar 21 13:57:46 likewise, gm Mar 21 13:57:55 hi fooks Mar 21 13:58:11 No network, other than usb-gadget. Mar 21 13:58:36 Don't think the module is loaded in the default build. Mar 21 13:58:40 oh :/ Mar 21 13:58:48 gm florian Mar 21 13:58:55 I think udev might be blowing away all the static device nodes. Mar 21 13:58:58 these fine gentlemen here might be of more assitance Mar 21 13:59:07 I am still a noob heh Mar 21 14:03:45 I see "Starting udev", shortly followed by "Caching udev devnodes", "Populating dev cache" then "root: mount: special device /dev/mtdblock2 does not exist" Mar 21 14:04:09 Oddly that node exists in the tarball version of the image Mar 21 14:07:59 Agh! My local network uses the same subnet as usb0 :-( Mar 21 14:29:10 that was your problem? Mar 21 14:39:12 03Mike Westerhof  07org.openembedded.dev * rd29c8d9fba 10openembedded.git/recipes/inetutils/inetutils_1.5.bb: Mar 21 14:39:13 inetutils: Copy in config.rpath prior to running configure task Mar 21 14:39:13 (fixes configure failure due to missing file so that inetutils builds again) Mar 21 15:02:12 03Koen Kooi  07org.openembedded.dev * r097e2a981e 10openembedded.git/conf/distro/angstrom-2008.1.conf: angstrom: bump version Mar 21 18:18:31 hmm Mar 21 18:19:16 JaMa: things being rebuilt unnecessarily was a bug with the base class split that was fixed on friday evening, as far as I know. Mar 21 19:05:11 http://gist.github.com/339495 - thoughts? Mar 21 19:05:19 python code wise Mar 21 19:09:59 hallo Mar 21 19:35:14 hmmm Mar 21 19:41:42 likewise: managed to get the hawk @ 456 Mhz, see the thread in oe ML for details; kudo's go to koen and ti Mar 21 19:46:30 eFfeM1: nice Mar 21 19:51:41 works like a charm Mar 21 20:03:45 eFfeM1: You part of the pixhawk team? Mar 21 20:06:57 nope Mar 21 20:08:11 aditya_1 it is a hawkboard Mar 21 20:08:22 www.hawkboard.org Mar 21 20:18:00 morning Mar 21 20:19:27 thats evening to you frenchie Mar 21 20:19:28 :P Mar 21 20:21:07 hi hrw Mar 21 20:21:50 ~ugt Mar 21 20:21:51 well, ugt is Universal Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Mar 21 20:22:51 pssh Mar 21 20:22:54 lies Mar 21 20:24:02 * kergoth <3 ugt Mar 21 20:24:20 heh Mar 21 20:24:36 fianlly got OE up and running again Mar 21 20:24:43 from the lesser known automake features department: since 1.10, you can put --install in ACLOCAL_AMFLAGS to get it to automatically copy the .m4's from the system macro dir into a local m4 dir Mar 21 20:30:30 morning kergoth Mar 21 20:32:13 hey pb__ Mar 21 20:34:50 pb__: not ideal, since it does end up returning createCopy'd datastores for each version, but..: http://gist.github.com/338382 Mar 21 20:35:17 http://github.com/kergoth/BitBake/tree/bbversions Mar 21 20:36:12 ah, cute Mar 21 20:37:31 need to profile it, to see what the impact of it is with large numbers of versions Mar 21 20:37:33 but it does work Mar 21 20:39:46 I wonder did someone read my mails about busyboxless rootfs Mar 21 20:41:22 * kergoth_ read it, but hasn't had a chance to reply Mar 21 20:41:29 seemed sane to me Mar 21 20:43:38 bye Mar 21 20:43:47 * sigh * Mar 21 20:43:55 half the things you people say goes completely over my head :/ Mar 21 20:50:55 hmm.. Mar 21 20:51:57 i wonder if i should add the original PV to overrides after the new one Mar 21 20:53:15 hi Mar 21 20:54:48 florian, can you help me out with netbook pro? I'm trying to build a boot image... I can make the header easily enough but am stuck with the kernel+initrd glue code Mar 21 20:55:23 hi EdCa229 Mar 21 20:55:35 :) Mar 21 20:55:58 oh... this was quite some time ago. Mar 21 20:56:36 i realise! I'd appreciate any pointers as currently it seems the only way forward is to dissasemble existing images Mar 21 20:56:38 I remember I wrote a description how this boot images are built and I remember somone built a tool to do this. Mar 21 20:56:55 theres some kind of bootstrap between the kernel and initrd we can't reproduce Mar 21 20:57:38 EdCa229: http://www.linuxtogo.org/~anonymous/netbook-booting.txt Mar 21 20:58:35 I have that tool... mk-nBkProOsImg... and the text file... but it's the bootstrap code I'm struggling with Mar 21 20:59:07 ah... there is acopy of this in the same dir anyway Mar 21 20:59:23 it's described in that text file, but not detailed... and the tool doesn't produce the bootstrap Mar 21 21:00:04 any ideas? Mar 21 21:00:27 * florian tries to remember Mar 21 21:03:38 EdCa229: ah... what you want to create is the nBkPro.img image, right? Mar 21 21:04:05 yep Mar 21 21:04:14 * kergoth wonders if bbversions is really worthwhile.. most people seem to want to ditch old versions of things, but there could be value.. hmm Mar 21 21:04:22 well, the nBkProOs.img Mar 21 21:05:31 it's got a header, some jump instruction, then kernel, some bootstrap code, then initrd Mar 21 21:05:42 the tool procudes the header Mar 21 21:05:53 and the kernel and initrd are easy Mar 21 21:06:01 buut the other bits.... *eek* Mar 21 21:13:24 hmm... I see. Mar 21 21:16:02 hrw, I have some patches at work for a busyboxless root Mar 21 21:16:55 Well, at the moment it replaces busybox with a list of depends. Mar 21 21:21:26 any clues? Mar 21 21:28:59 ah, the famous boost header Mar 21 21:31:13 I think you are probably best off just treating most of those 256 bytes as being magic bits that you need to copy verbatim from a working image. Mar 21 21:33:39 * pb__ dinner time now Mar 21 21:36:32 apart from the checksum maybe ;) Mar 21 21:37:25 florian, got logged out... did i miss something? Mar 21 21:37:46 EdCa228: you did... maybe Mar 21 21:38:03 :/ Mar 21 21:39:14 pb__ mentioned that you might just want to take most of the bits as given magic :) Mar 21 21:40:55 well... I would if I can get them to make me a bootable image! but it isn't just a matter of using dd to strip those bits and put them in a new file... that doesn't work Mar 21 21:41:35 presumably they are dependent on the size of the kernel and maybe other things Mar 21 21:41:37 :( Mar 21 21:42:08 kergoth, I don't think most people want to dump old versions, at least without concious thinking Mar 21 21:42:33 I've been thinking that most metadata isnt really bound to one version Mar 21 21:42:37 dumping recipes just because they are old is potentially annoying to their users Mar 21 21:42:40 its bound to a range of versions, based on when changes occurred upstream Mar 21 21:42:51 so if you could have a recipe for each range, you could have every version buildable Mar 21 21:43:12 interesting Mar 21 21:44:00 that would be helpful for things like boost that have the occasional api change on minor version number changes Mar 21 21:45:21 Crofton: http://gist.github.com/338382, http://github.com/kergoth/BitBake/commits/bbversions Mar 21 21:45:30 I guess that means you have no idea though? ...you can't dredge up some long lost memory (or even better code or notes) from somewhere? Mar 21 22:05:29 EdCa228: not really... might be a bug in the existing tool. but since the important bits are basically documented and we know some working values it should be possible to get this fixed. Mar 21 22:08:07 Let me see if I can find more about the wrapper Mar 21 22:11:01 :) Mar 21 22:12:54 I wonder if you could use U-Boot instead of this wrapper... Mar 21 22:14:39 ive no idea... presumably u-boot would have to be flashed on there somehow Mar 21 22:15:48 i considered using haret from win ce but ive partitioned my nand and dont know how to revert it... and i dont want to play with too much in case i erase boos and then wont be able to boot anything! Mar 21 22:22:11 brb Mar 21 22:23:20 EdCa22: in fact this wrapper is a simple piece of assembler code. what it does is mostly what a simple bootloader would do. Mar 21 22:24:12 thought it was something like that... Mar 21 22:24:26 any idea how I can modify it per image? Mar 21 22:24:39 using u-boot might be sort of overkill but it might be easy to do Mar 21 22:25:24 would you boot u-boot from boost or install it in nand? Mar 21 22:25:56 florian: I think the crc is just for the header, not the whole image. but, iirc, there is an image size field in there. Mar 21 22:26:32 so, annoyingly, you'd need to adjust the image size and then recalculate the crc whenever the kernel or rootfs changes size Mar 21 22:26:34 pb__: yes this might be correct... the text says something about the header too. Mar 21 22:27:14 yeah... the header is easy enough to reproduce Mar 21 22:27:51 in fact the tool does that part for you... pb you are correct on both counts Mar 21 22:28:28 i have pulled the two existing working images to pieces completely Mar 21 22:29:01 The wrapper does not have to be changed for each image - it takes two parameters at build time: kernel and initrd position Mar 21 22:30:36 ahh.. I need to know how to change those parameters and what to! Mar 21 22:30:50 the header does change every time though Mar 21 22:32:51 the header contains 2 crcs, one for the image and one for the header. and an image size field Mar 21 22:34:13 wait.. if the wrapper doesnt need to be changed then why can't I make an working image? Mar 21 22:35:19 what I have done is pull out those parts of the code, and then jam them together before the kernel and before the initrd Mar 21 22:35:39 but in the two working examples I have seen those parts are slightly different in places Mar 21 22:36:04 maybe 3 or 4 words are different Mar 21 22:40:15 florian... does that mean you have code for the wrapper? Mar 21 22:42:06 * florian tries to fin out how this used to work Mar 21 22:42:32 EdCa22: That's why some people here are able to say what it does Mar 21 22:43:36 this whole lowlevel stuff is a terrible collection... who coded all this?! Mar 21 22:47:06 EdCa22: I see why this is that complicated... kernel and initrd are included as a binary section into the assembler file Mar 21 22:47:18 so it needs to be changed Mar 21 22:48:17 what it makes harder is the fact that the whole thing after the header is compressed Mar 21 22:48:19 yeah... it's lame. ive been working with dd, zpipe and a vbindiff for days trying to build something that works Mar 21 22:48:35 the header isn't compressed Mar 21 22:48:40 right Mar 21 22:48:57 but the binary after the header is Mar 21 22:48:58 but you can decompress the rest by cutting off the header and running it through zpipe Mar 21 22:49:46 then i find a single word, an address i think, followed by the uncompressed kernel, ~1052 bytes of bootstrap with kernel parameters etc, and then the initrd Mar 21 22:50:48 right - that's the wrapper. the kernel parameters are compiled in there Mar 21 22:50:51 wait, sorry, after the header is an instruction and an address i think, then the kernel Mar 21 22:51:09 yep Mar 21 22:51:41 but i just don't know how to make the wrapper Mar 21 22:51:59 i think the initial instruction makes execution jump to the wrapper Mar 21 22:52:15 quite likely yes Mar 21 22:53:02 but i don't even know how that jump address is derived Mar 21 22:53:26 That's up to gcc I guess... Mar 21 22:54:06 I wonder if it would be a good idea just to have a U-Boot binary instead of the kernel+initrd. This would be able to deal with Linux then. Mar 21 22:54:25 The only thing you's have to take care about is Boost itself Mar 21 22:55:40 i don't know how I'd go about that.... i don't know that much about uboot Mar 21 22:55:49 just put uboot and a header? Mar 21 22:56:48 uboot would then need drivers for the cf to get a kernel+initrd... and that means work unless it's a standard implementation for pxa255 Mar 21 22:56:58 unless if it supports serial loading Mar 21 22:57:36 but i mean, maybe uboot even would need some kind of bootstrap for boost? Mar 21 22:58:22 you can configure it to act as a secondary stage bootloader. so this might just do what we ned. Mar 21 23:02:22 but wait... presumably at some point you were able to build images for it... I guess that code is lost in the tubes somewhere? Mar 21 23:07:03 so if i build uboot as a 2nd stage bootloader and put a header on it it should just work? Mar 21 23:14:18 Its quite likely I think... but not sure of course. I never tried this myself. Mar 21 23:21:20 hmmm... i feel so close getting this working as it is. but if u-boot can work like that i'll give it a shot Mar 22 00:03:52 EdCa22: let me know if it works! Mar 22 00:04:06 thanks for your help florian Mar 22 00:04:10 I will be in touch Mar 22 00:12:37 EdCa22: ok, great Mar 22 00:12:43 * florian -> sleep, good night Mar 22 00:12:53 ;) Mar 22 01:33:14 03Christopher Larson  07org.openembedded.dev * r0263048328 10openembedded.git/classes/ (base.bbclass utils.bbclass): Mar 22 01:33:14 base.bbclass: add popen/system convenience functions Mar 22 01:33:14 Provides oe_popen, which is a subprocess.Popen wrapper that automatically Mar 22 01:33:14 provides our exported variables in the environment, including the PATH, and Mar 22 01:33:14 oe_system, which is just a wrapper that acts like system. Mar 22 01:33:15 Signed-off-by: Chris Larson Mar 22 01:33:16 Acked-by: Tom Rini Mar 22 01:33:24 03Chris Larson  07org.openembedded.dev * rd91b6bc423 10openembedded.git/classes/ (src_distribute.bbclass src_distribute_local.bbclass): Mar 22 01:33:25 src_distribute: simplify do_distribute_sources Mar 22 01:33:25 Signed-off-by: Chris Larson Mar 22 02:09:10 yarr **** ENDING LOGGING AT Mon Mar 22 03:00:02 2010