**** BEGIN LOGGING AT Mon Oct 26 02:59:58 2009 Oct 26 06:42:18 how do i know if linux overwrites the u-boot setttings for gpio and mux? Oct 26 06:42:30 im trying to toggle a pin Oct 26 06:42:55 through a kernel space module(just board-overo) Oct 26 07:03:14 what is CONFIG_OMAP_GPIO_SWITCH in defconfig Oct 26 07:03:33 its set to no, is this why i cant toggle gpio from kernel space? Oct 26 08:14:35 03Michael Smith  07koen/static-libs-rework * re698a27699 10openembedded.git/ (conf/checksums.ini recipes/libsdl/libsdl-mixer_1.2.9.bb): Oct 26 08:14:35 libsdl-mixer: add 1.2.9 Oct 26 08:14:35 Signed-off-by: Michael Smith Oct 26 08:14:37 03Michael Smith  07koen/static-libs-rework * r3b408bf7c8 10openembedded.git/recipes/curl/ (7 files): Oct 26 08:14:37 curl: support openssl in CURL_FEATURES. Oct 26 08:14:40 Default is still gnutls. Oct 26 08:14:42 Also switch to INC_PR. Oct 26 08:14:44 Signed-off-by: Michael Smith Oct 26 08:14:46 03Koen Kooi  07koen/static-libs-rework * ra6a6e308c7 10openembedded.git/classes/package.bbclass: package bbclass: don't strip static libs so aggressively Oct 26 08:14:49 03Koen Kooi  07koen/static-libs-rework * re8d991f7c3 10openembedded.git/recipes/glibc/glibc-package.bbclass: (log message trimmed) Oct 26 08:14:52 glibc package bbclass: tweak library packaging Oct 26 08:14:54 libcidn and libmemusage have there own package now Oct 26 08:14:56 Static libs move around a bit: Oct 26 08:14:58 glibc-dev: Oct 26 08:15:00 /usr/lib/libpthread_nonshared.a Oct 26 08:15:02 /usr/lib/libc_nonshared.a Oct 26 08:15:04 03Koen Kooi  07koen/static-libs-rework * r5471a51fbf 10openembedded.git/recipes/libxml/ (libxml2.inc libxml2_2.7.3.bb): libxml2: stop overriding PACKAGES Oct 26 08:15:13 03Koen Kooi  07koen/static-libs-rework * rf899c835e8 10openembedded.git/conf/bitbake.conf: bitbake conf: put static libs in ${PN}-static Oct 26 08:15:16 03Koen Kooi  07koen/static-libs-rework * rf6e6b0d59e 10openembedded.git/recipes/bison/bison_2.3.bb: Oct 26 08:15:19 bison: put liby.a in ${PN} Oct 26 08:15:21 Phil Blundell said: Oct 26 08:15:23 >> bison-2.3-r0: /usr/lib/liby.a Oct 26 08:15:27 That one is a special case: it wants to stay in the main bison package, Oct 26 08:15:29 since bison itself is a development tool. I'm not sure why it is a Oct 26 08:15:31 static-only library; that might be an error. Oct 26 08:15:33 03Michael Smith  07koen/static-libs-rework * r388a34e028 10openembedded.git/ (3 files in 2 dirs): Oct 26 08:15:36 curl, curl-native: add 7.19.6 Oct 26 08:15:38 Signed-off-by: Michael Smith Oct 26 08:15:40 Acked-by: Khem Raj Oct 26 08:15:46 03Koen Kooi  07koen/static-libs-rework * r5bc902807f 10openembedded.git/conf/distro/include/ (angstrom-eglibc.inc angstrom-glibc.inc): angstrom: change *glibc build to -Os Oct 26 08:15:49 03Koen Kooi  07koen/static-libs-rework * r7b1c060e85 10openembedded.git/recipes/flex/ (flex.inc flex_2.5.31.bb): (log message trimmed) Oct 26 08:15:52 flex: put libfl into ${PN} Oct 26 08:16:00 Phil Blundell said: Oct 26 08:16:02 > > flex-2.5.31-r4: /usr/lib/libfl.a Oct 26 08:16:04 This is like bison. Oct 26 08:16:06 >> bison-2.3-r0: /usr/lib/liby.a Oct 26 08:16:08 That one is a special case: it wants to stay in the main bison package, Oct 26 08:39:34 gm Oct 26 08:51:27 03Jeremy Lainé  07org.openembedded.dev * r1f7a632a96 10openembedded.git/classes/update-rc.d.bbclass: Oct 26 08:51:27 update-rc.d.bbclass: make prerm and postrm dpkg-compatible Oct 26 08:51:27 * prerm : only stop service at the "upgrade" and "remove" steps Oct 26 08:51:27 * postrm : only remove symlinks at the "remove" and "purge" steps, after Oct 26 08:51:27 checking the init script no longer exists Oct 26 08:55:39 hello Oct 26 08:55:49 anybody using the ntpdate package here? Oct 26 08:56:35 good morning Oct 26 08:56:45 jeremy_laine: yes, but never tried in OE Oct 26 09:32:33 hi all Oct 26 09:45:49 hi florian Oct 26 09:54:22 mckoan: heh, sorry for the lag Oct 26 09:54:48 mckoan: I was indeed refering to the OE version, I'm not too happy with it's init script Oct 26 09:55:19 mckoan: I don't think it makes a whole lot of sense to run ntpdate at startup as there's no garranty you have network access Oct 26 09:55:27 florian: good morning Oct 26 10:05:07 jeremy_laine: sounds like a good candidate to put in network post-up scripts Oct 26 10:05:45 yes, that's what debian does nowadays Oct 26 10:06:08 running ntpdate from /etc/init.d is indeed a bit too retro and doesn't make a lot of sense in the 21st century. Oct 26 10:06:42 hi lrg Oct 26 10:08:54 hey florian Oct 26 10:09:32 pb_: that's a bit harsh but very true :) Oct 26 10:13:33 jeremy_laine, pb_: is there a "we have internet access" event in Linux? dhcp and avahi and fixed configuration clients all seem to do their own stuff. Oct 26 10:14:45 jeremy_laine, pb_: I remember one particular case where getting the DHCP lease resulted in the NTP server getting known, the time jumping, which then resulted in the DHCP lease being immediately "expired", resulting in release, resulting in rootfs/nfs loss... Oct 26 10:15:26 i.e. the time jump made the client skip the "renew" fase and go into "release/discover" phase (as per the DHCP spec). Oct 26 10:15:56 A real 21st solution is a tighter integration of the state machines. Oct 26 10:15:57 anybody was able to use mouse in qemu x86 image ? Oct 26 10:17:09 likewise: no, I think this (and indeed the general state of the art in linux networking) is still full of suck. Oct 26 10:17:44 likewise: NetworkManager does produce such an event if you're using that. Oct 26 10:18:10 there's the NetworkManager, of course, which I believe is meant to solve this problem, but it doesn't seem to do so perfectly. Oct 26 10:18:25 Which is probably a good enough standard for end node apps. Oct 26 10:18:42 pb__: To be fair to NM, half the trouble it has is the massive amount of suck in the wireless drivers. Oct 26 10:19:06 yeah, that is probably true. Oct 26 10:19:06 agreed on all points, but isn't the NM a GUI tool? Oct 26 10:19:17 It's a DBus service. Oct 26 10:19:21 likewise: no, NM itself is a dbus backend Oct 26 10:19:23 with GUI front ends. Oct 26 10:19:27 k Oct 26 11:11:30 hi, is there a standard way to create an archive with the current toolchain in OE, I can well pack tmp/cross but I wonder if there is a conventional naming scheme for toolchain archives Oct 26 11:11:54 s/,/? Oct 26 11:14:17 03Michael Smith  07org.openembedded.dev * rdba8dec409 10openembedded.git/classes/gitver.bbclass: (log message trimmed) Oct 26 11:14:17 gitver.bbclass: fix git dependency tracking Oct 26 11:14:17 .git/HEAD doesn't usually change on a commit, but the ref it points to Oct 26 11:14:17 (e.g. .git/refs/heads/master) should. Handle this and another couple of Oct 26 11:14:17 cases: if a tag is added without a new commit, or if the ref is in Oct 26 11:14:21 packed-refs. Oct 26 11:14:23 Signed-off-by: Michael Smith Oct 26 11:20:19 03Frans Meulenbroeks  07org.openembedded.dev * r4882a87cc5 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Oct 26 11:20:22 03Frans Meulenbroeks  07org.openembedded.dev * r8afa9e311d 10openembedded.git/ (conf/checksums.ini recipes/tgt/tgt_0.9.10.bb): tgt (Linux SCSI target framework) : recipe created Oct 26 12:13:17 morning Oct 26 12:13:24 florian, we need a section at OEDEM to appoint people to do all the things no one wants to do :) Oct 26 12:13:29 hrw, gm Oct 26 12:13:52 hrw, would it be a good idea to have a erport on OE at LEC during OEDEM? Oct 26 12:14:22 Crofton|work: report about OE on ELC-E during OEDEM? Oct 26 12:14:53 hi hrw Oct 26 12:15:11 yeah Oct 26 12:15:14 Crofton|work: heh... indeed Oct 26 12:16:14 http://tree.celinuxforum.org/CelfPubWiki/ELCEurope2009Presentations?action=AttachFile&do=view&target=Hombourger-Why_oe_good_foundation_for_mv.pdf Oct 26 12:16:26 cool Oct 26 12:16:34 can anyone shed some light on why PR would be silently overriden to a blank string? Oct 26 12:16:36 we already know that :) Oct 26 12:16:47 only seems to affect machine architecture recipes Oct 26 12:16:50 I am curious what you thought of people impressions Oct 26 12:18:30 hrw: Yes... read this half an hour ago. Yes this is interesting :) Oct 26 12:18:59 mike_cw: which recipes, specifically? Oct 26 12:19:04 Alex de Vries had nice talk but did not uploaded slides yet ;( Oct 26 12:19:15 at a first guess, you may have fallen victim to MACHINE_KERNEL_PR. Oct 26 12:19:33 pb__, they are custom ones in an overlay. armv5te ones in the same overlay work as expected. Oct 26 12:19:45 pb__: M_K_PR should not change anything for devices which do not use it Oct 26 12:24:02 Here is one of the recipes http://pastebin.com/d5e8d0b8d Oct 26 12:24:25 but it is pretty simple, and it does actually work - just the package has no PR in the filename Oct 26 12:25:18 hi, is there a standard way to create an archive with the current toolchain in OE, for external distribution? I can well pack tmp/cross but I wonder if there is an automation to do that and/or a conventional naming scheme for toolchain archives. Oct 26 12:25:18 ao2: bitbake meta-toolchain Oct 26 12:25:32 hrw, thanks Oct 26 12:28:44 fosdem devroom call is up Oct 26 12:29:08 it seems like the embedded room has been following the "new" devroom model all along :) Oct 26 12:32:17 mike_cw: if you take out the "inherit module", does that cause your PR to come out correctly? Oct 26 12:32:26 obviously the recipe won't actually work under those conditions, but you can check with bitbake -e Oct 26 12:35:07 pb__, just running that. i guess it will work because i can see PR is forced in module-base.bbclass Oct 26 12:35:49 yes. that does make it come out correclty. i'll move the inherit before the PR in the recipe Oct 26 12:56:03 pb__: i had to clean build the kernel for some reason, but that has all run through correctly now. thanks for the pointers Oct 26 12:58:52 03Xerxes Rånby  07org.openembedded.dev * r66c4fef12c 10openembedded.git/ (5 files in 2 dirs): Oct 26 12:58:52 llvm2.6: Updated to use llvm 2.6 release tarball. Oct 26 12:58:52 llvm2.6-native: Likewise. Oct 26 12:58:52 checksums.ini: add sums for llvm-2.6.tar.gz Oct 26 13:17:42 hrw: what was Alex de Vries' talk about? Oct 26 13:19:54 anyone an idea what is wrong in the recipe I am crafting if I get this during the build: Oct 26 13:19:56 arm-angstrom-linux-gnueabi-ld: unrecognized option '-Wl,-rpath-link,/home/frans/oe/tmp_angstrom/staging/armv7a-angstrom-linux-gnueabi/usr/lib' Oct 26 13:20:55 eFfeM: ${LD} Oct 26 13:21:08 likewise: about buildsystems, which talks were at elc-e about them, what do they do, what they miss Oct 26 13:22:46 pb__, thanks, guess this means I am in the shit, this recipe builds both an exe and a module (trying to port open-iscsi) Oct 26 13:23:26 eFfeM: yeah, in that case I think you are in a bad place. the best way to deal with that is to split it into two recipes. Oct 26 13:24:28 under some conditions (e.g. KERNEL_CCSUFFIX set) you need to use a completely different compiler for kernel modules than for regular exes. Oct 26 13:25:02 although it is possible in theory to make a single recipe that will handle both, it will be complicated and fragile and difficult to get right. Oct 26 13:25:04 eFfeM: madwifi also does this, and nvidia driver as well. (indeed, those are shitty recipes) Oct 26 13:26:36 ah ok, i think i can easily split Oct 26 13:29:19 see lirc for example, which I think is split like that. Oct 26 13:29:21 decided earlier today that it was a nice day to learn about iscsi, not so sure any more ;-) Oct 26 13:29:25 heh Oct 26 13:29:30 i've made two recipes Oct 26 13:29:45 top level makefile already split out a user and kernel section Oct 26 13:29:54 * hrw -> food Oct 26 13:31:26 hrw, enjoy :-) Oct 26 13:39:25 thx Oct 26 13:39:58 makiing progress, user part now compiles but during package_write_ipk I get: Error: CONTROL/control is missing field Description Oct 26 13:40:11 never seen this before, anyone a clue? Oct 26 13:40:24 deb packaging? Oct 26 13:40:33 eFfeM: empty DESCRIPTION field? Oct 26 13:40:39 yessss Oct 26 13:40:50 decided to fill in the headers later Oct 26 13:40:59 didn't expect this error from it Oct 26 13:43:26 now the user part builds, hrw, thanks; on to the kernel challenge Oct 26 13:51:48 hrw: hi Oct 26 13:52:43 hrw: apart from obscure bugs, why shouldn't c7x0 boot from nand but only after kexec with 2.6.31? Oct 26 13:53:23 of course all mtd/mtdbloxk is compiled in kernel Oct 26 13:53:53 anyone can boot an armv5te from nand with 2.6.31? Oct 26 13:54:02 confirmations? Oct 26 13:55:01 ah, mtd partitions are regularly found after boot from SD/CF Oct 26 13:57:32 ant_work: I did not used mtd on armv5te with 2.6.31 at all yet Oct 26 13:57:59 ant_work: and I did not used c760 for ~year Oct 26 13:58:44 I'll try one step back..2.6.30 Oct 26 13:59:02 btw, our 2.6.31 is not yet the latest? Oct 26 13:59:10 have you committed r5? Oct 26 13:59:29 pb__ tried to build with oe_runmake 'KSRC=${STAGING_KERNEL_DIR}' 'LD=${KERNEL_LD}' but still get the linker error Oct 26 13:59:41 any idea ? Oct 26 14:00:36 -Wl must be given to ${CC}, not ${LD} directly AFAIK Oct 26 14:00:56 ah ok, will look at that Oct 26 14:01:02 -Wl means "pass to linker" Oct 26 14:02:50 btw not sure if I am happy that if you intherit module that the recipe uses MACHINE_KERNEL_PR as number and ignores PR, Oct 26 14:03:23 ant_work: no, my linux updates are not merged Oct 26 14:04:04 that means that if I have a recipe like the one I am crafting now, (open-iscsi-kernel) that if the recipe changes all MACHINE_KERNEL_PR's should be bumped (that is: if I understood things correctly) Oct 26 14:05:49 hmmm, MACHINE_KERNEL_PR * 100 + PR Oct 26 14:24:52 is it possible to have opkg prioritise a local repository over the Angstrom one, regardless of version? Oct 26 14:25:01 iirc nope Oct 26 14:25:11 hmm Oct 26 14:25:55 it would be nice if there was something analogous to bitbake collections Oct 26 14:29:53 patches welcome? Oct 26 14:30:13 eFfeM: I think you need to do something with LDFLAGS as well in the kernel case. module_base.bbclass should take care of it, I suggest you just inherit that class. Oct 26 14:30:22 regular ${LDFLAGS} is not good news for kernel modules Oct 26 14:59:32 03Steffen Sledz  07org.openembedded.dev * r249a10ce18 10openembedded.git/conf/checksums.ini: ltrace_0.5.3: missing checksum added Oct 26 15:11:23 hi Oct 26 15:11:42 does someone have any experience using ubifs? Oct 26 15:11:54 +with Oct 26 15:15:32 pb__ was afk, I already set LDFLAGS to "" and it compiles and links Oct 26 15:15:38 still need to fix install Oct 26 15:21:08 pb__: hi Oct 26 15:21:25 pb__: I'm looking at 3 recipes giving QA about packaging Oct 26 15:21:35 the old .so in non-dev issue Oct 26 15:22:07 03Steffen Sledz  07org.openembedded.dev * r93443046f1 10openembedded.git/conf/checksums.ini: ltrace_0.5.3: some more missing checksums added Oct 26 15:22:21 the packages are libopieboex0 libgtktilus and libpcre Oct 26 15:31:16 03Xerxes Rånby  07org.openembedded.dev * rb39a67f8b6 10openembedded.git/conf/checksums.ini: llvm2.6: fix checksum Oct 26 15:39:12 pb__: e.g. libopieobex0 uses oe_libinstall -so libopieobex ${D}${palmtopdir}/plugins/obex/ Oct 26 15:39:30 and FILES_${PN} = "${palmtopdir}/plugins/obex/ ${palmtopdir}/pics/obex/" Oct 26 15:40:06 guys: I have new machine to add (ST NHK-15 devboard). It uses own copy of kdrive 1.3.0.0 recipe. Oct 26 15:40:25 what would be better: ship recipe or merge into OE one + some tweaks inside? Oct 26 15:40:57 hrw: usually i would say merging is the better idea... Oct 26 15:41:12 hrw: what did you change? Oct 26 15:41:45 they have two own patches which contains also some of OE ones ;( Oct 26 15:42:29 pb__: what happens is that the .so is not considerated 'dev' because is out-of-path Oct 26 15:43:05 hrw: are the remaining ones necessary? ;) Oct 26 15:43:53 03Frans Meulenbroeks  07org.openembedded.dev * r2683426742 10openembedded.git/ (3 files in 2 dirs): open-iscsi: added recipe; still needs stress-testing Oct 26 15:44:02 ok, there's the open-iscsi recipes. not too heavy tested but at least it is now archived Oct 26 15:44:04 03Frans Meulenbroeks  07org.openembedded.dev * rf8e861e80c 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Oct 26 15:44:04 03Frans Meulenbroeks  07org.openembedded.dev * rd7e9f09e29 10openembedded.git/recipes/tgt/tgt_0.9.10.bb: tgt: updaded recipe to use ${PV} for download Oct 26 15:44:19 florian: HW accelerated X11 is good thing Oct 26 15:44:36 hrw: I agree :) Oct 26 15:44:38 hrw: is that the last kdrive working on w100 Oct 26 15:44:40 ? Oct 26 15:45:01 I made two recipes: open-iscsi-kernel and open-iscsi-user; actually these should go together, is there a way to make something like a dummy package that depends on both ? Oct 26 15:45:06 03Florian Boor  07org.openembedded.dev * r7ef1593130 10openembedded.git/recipes/prelink/ (prelink_20061027.bb prelink_20071009.bb): prelink: Change postrm script to prerm. Calling 'prelink' after removing the binary is likely to fail. Oct 26 15:46:18 ant_work: 1.2.0 was last iirc Oct 26 15:46:38 :/ Oct 26 15:50:37 ant_work: right, you would need to add the .so path to FILES_${PN}-dev explicitly Oct 26 15:51:07 or, perhaps better in this case, just delete it: the .so file for a plugin is pretty useless Oct 26 15:52:20 pb__: btw -dbg is wrong too Oct 26 15:52:45 yeah, same applies Oct 26 15:53:09 it might be possible to patch opie.bbclass to add ${palmtopdir}/plugins/.debug to FILES_${PN}-dbg globally Oct 26 15:54:05 pb__: yea, that was the underlying question :) Oct 26 16:02:07 eFfeM: make the -user package RRECOMMEND -kernel ? Oct 26 16:02:41 eFfeM: the way we typically do not enforce RRECOMMENDS AFAIK Oct 26 16:02:58 s/the way/the other way/ Oct 26 16:05:57 03Frans Meulenbroeks  07org.openembedded.dev * rc35a7aa427 10openembedded.git/recipes/open-iscsi/open-iscsi-kernel_2.0-871.bb: open-iscsi: added MACHINE_KERNEL_PR_append = "a" Oct 26 16:09:08 likewise, thanks, there you go: Oct 26 16:09:11 03Frans Meulenbroeks  07org.openembedded.dev * r1164525915 10openembedded.git/recipes/open-iscsi/open-iscsi-user_2.0-871.bb: open-iscsi-user: added RRECOMMENDS Oct 26 16:09:18 03Marcin Juszkiewicz  07org.openembedded.dev * r703ba278b7 10openembedded.git/conf/machine/nhk15.conf: nhk15: added machine config for ST Microelectronics NHK15 devboard Oct 26 16:09:19 03Marcin Juszkiewicz  07org.openembedded.dev * ref40258f38 10openembedded.git/recipes/linux/ (10 files in 2 dirs): Oct 26 16:09:19 linux: added NHK15 support for 2.6.20 version Oct 26 16:09:19 staged headers are needed for xserver-kdrive-nomadik X11 acceleration Oct 26 16:09:20 03Marcin Juszkiewicz  07org.openembedded.dev * r0cf49e7dc8 10openembedded.git/recipes/xorg-xserver/ (3 files in 2 dirs): Oct 26 16:09:23 xserver-kdrive-nomadik: added 1.3.0.0 version Oct 26 16:09:25 This version has hardware acceleration working. Oct 26 16:09:27 NOTE: patches were applied and refreshed - thats why they differ from ST Oct 26 16:09:29 ones. Oct 26 16:12:12 hmm.. looks like I broke something Oct 26 16:13:21 03Klaus Kurzmann  07shr/import * ra0d1f55851 10openembedded.git/conf/distro/shr.conf: Oct 26 16:13:21 shr.conf: set provider for the new logo splash instead of the simple one Oct 26 16:13:21 Signed-off-by: Klaus Kurzmann Oct 26 16:13:22 03Klaus Kurzmann  07shr/import * r7f9e2cc94d 10openembedded.git/recipes/tasks/task-shr-minimal.bb: Oct 26 16:13:22 task-shr-minimal.bb: exchange ophonekitd with new phoneui and add fso-abyss Oct 26 16:13:25 Signed-off-by: Klaus Kurzmann Oct 26 16:13:27 03Klaus Kurzmann  07shr/import * recdcf40ea1 10openembedded.git/recipes/shr/phoneui-apps_git.bb: Oct 26 16:13:30 phoneui-apps_git.bb: make the main package an empty meta package Oct 26 16:13:32 Signed-off-by: Klaus Kurzmann Oct 26 16:13:32 cgit does not show my nhk15 changes, I cannot pull them Oct 26 16:14:24 eFfeM: I *think* that is the right approach, but I should look at other similar packages (we do not have many I think). Oct 26 16:16:34 03Sebastian Krzyszkowiak  07shr/import * r623ff70d0e 10openembedded.git/recipes/freesmartphone/ (2 files in 2 dirs): Oct 26 16:16:34 shr/merge: sync frameworkd recipe with shr/import and add patch for opimd support in oeventsd for SHR distro Oct 26 16:16:34 Signed-off-by: Klaus Kurzmann Oct 26 16:21:22 likewise: RRECOMMENDS is definitely better than nothing, it might even have to me RDEPENDS (if that exists, I do not really have a good mental overview of all things that can show up in a recipe). Oct 26 16:24:24 RDEPENDS is only appropriate if (a) the depending package would be totally useless without the depended-on one, and (b) the depended-one one is always guaranteed to exist as a package (which, in the case of kernel modules, means that it can't ever be built into the kernel statically). Oct 26 16:24:34 if either of those is not true, RRECOMMENDS would be better Oct 26 16:24:46 or, as you suggest, a third package which depends on both of the existing ones Oct 26 16:26:57 03Marcin Juszkiewicz  07org.openembedded.dev * r63fd61ae8a 10openembedded.git/conf/machine/nhk15.conf: nhk15: added machine config for ST Microelectronics NHK15 devboard Oct 26 16:26:58 03Marcin Juszkiewicz  07org.openembedded.dev * r6cb806430f 10openembedded.git/recipes/linux/ (10 files in 2 dirs): Oct 26 16:26:58 linux: added NHK15 support for 2.6.20 version Oct 26 16:26:58 staged headers are needed for x11 acceleration Oct 26 16:27:00 03Marcin Juszkiewicz  07org.openembedded.dev * r083679cd96 10openembedded.git/recipes/xorg-xserver/ (3 files in 2 dirs): Oct 26 16:27:03 xserver-kdrive-nomadik: added 1.3.0.0 version Oct 26 16:27:05 This version has hardware acceleration working. Oct 26 16:27:07 NOTE: patches were applied and refreshed - thats why they differ from ST Oct 26 16:27:09 ones. Oct 26 16:27:11 first time I saw negative times in cgit ;d Oct 26 16:30:36 bye Oct 26 16:39:41 bbl Oct 26 17:04:36 pb__ i did RRECOMMENDS, but actually the user package is useless w/o the kernel module and the kernel module does nnot exist yet in the kernel tree, so it would be tricky to build it into the kernel Oct 26 17:05:41 pb__ if I create a 3rd package then how would I need to to that ? there is not really a deliverable so no .ipk generated it becomes sort of a virtual Oct 26 17:08:26 hi all, quick question Oct 26 17:08:43 I've just changed toolchain from the OE built one, to an external once Oct 26 17:08:46 s/once/one Oct 26 17:09:04 What command shall I use to clean everything & rebuild? Oct 26 17:09:11 bitbake -c clean world? Oct 26 17:09:30 morning Oct 26 17:09:55 greyback, i'm not sure it is recommendable to use an external toolchain Oct 26 17:10:06 but if you insist the best thing is to remove the tmp dir Oct 26 17:10:28 eFfeM: maybe not, but the toolchain OE builds doesn't work for me, so I'm trying a crosstool-ng one I made Oct 26 17:11:04 hm, why won't it work are you on an exotic target or on an exotic host Oct 26 17:11:40 I am rather confused about it. It's not exotic, an armv5te Oct 26 17:12:06 I expected the OE chain to just work, but no luck so far Oct 26 17:12:14 well the toolchain works for armv5te, I've build the kernel and console-image for sheevaplug this weekend Oct 26 17:12:23 and that is kirkwood which is armv5te Oct 26 17:12:32 That's what I would expect, hence my confusion Oct 26 17:12:52 guess there is something else wrong Oct 26 17:13:06 not sure what as you didn't tell what your problem was Oct 26 17:13:18 btw make sure you do not have symlinks in your path, that might cause trouble Oct 26 17:13:32 hmm, that's news to me Oct 26 17:13:54 no, not that. Oct 26 17:14:03 I'm trying to get OE working for the Neuros OSD Oct 26 17:14:15 It's already defined as a machine, with a distro there too Oct 26 17:14:34 But the kernel is builds fails to boot completely Oct 26 17:14:59 No output at all after uboot, even with debugging on Oct 26 17:15:04 greyback: http://wiki.openembedded.net/index.php/Getting_started Oct 26 17:15:05 You must choose a location with no symlinks above it Oct 26 17:15:22 eFfeM: yeah, I checked that, that's not my problem Oct 26 17:15:52 greyback: typically this is an indication that your u-boot settings are wrong Oct 26 17:16:09 or that it got improperly flashed or loaded or whatever you do to get it on the board Oct 26 17:16:15 it boots the same kernel just fine, when built with different toolchain Oct 26 17:16:16 does u-boot say kernel loading ? Oct 26 17:16:27 yes, I'm loading over tftp Oct 26 17:16:38 "Starting linux..." is the last message I get Oct 26 17:17:18 morning all Oct 26 17:17:58 yes, had this myself in the past as well guess your start address is not ok, not sure if the kernel can be loaded at any address eiterh Oct 26 17:19:09 I suspected this, but with same kernel andsame uboot config, the toolchain was only differ Oct 26 17:19:21 Will investigate my uboot conf a little more closely Oct 26 17:24:03 gl Oct 26 17:26:04 03Gregoire Gentil  07org.openembedded.dev * rb293c57085 10openembedded.git/recipes/x-load/ (3 files in 2 dirs): x-load: sync touchbook patches with AI repo Oct 26 17:26:04 03Gregoire Gentil  07org.openembedded.dev * r6075c66b7d 10openembedded.git/recipes/linux/ (7 files in 2 dirs): linux-omap 2.6.29: sync touchbook patches with AI repo Oct 26 17:26:05 03Gregoire Gentil  07org.openembedded.dev * r27e7c4d855 10openembedded.git/recipes/perl/ (perl-5.8.8/Makefile.SH.patch perl_5.8.8.bb): perl 5.8.8: pass unidatafiles as extra arg to miniperl Oct 26 17:26:06 03Gregoire Gentil  07org.openembedded.dev * r995ee4b0a9 10openembedded.git/conf/checksums.ini: checksums: merge checksums with AI repo Oct 26 17:26:08 03Gregoire Gentil  07org.openembedded.dev * ra466778b6a 10openembedded.git/recipes/u-boot/ (u-boot-git/omap3-touchbook/new-pinmux.patch u-boot_git.bb): u-boot git: sync touchbook patches with AI repo Oct 26 17:26:11 03Gregoire Gentil  07org.openembedded.dev * r353fdfb2a2 10openembedded.git/ (8 files in 2 dirs): (log message trimmed) Oct 26 17:26:14 perl: sync extra perl recipes with AI repo Oct 26 17:26:16 libcompress-raw-zlib-perl: add 2.015 Oct 26 17:26:18 libcrypt-rijndael-perl: add 1.07 Oct 26 17:26:20 libdigest-sha: add 5.47 Oct 26 17:26:22 libio-compress-base-perl: add 2.015 Oct 26 17:26:24 libio-compress-zlib-perl: add 2.015 Oct 26 17:26:26 03Koen Kooi  07org.openembedded.dev * rced897b174 10openembedded.git/recipes/gnome/ (gnome-panel-2.28.0/idl-sysroot.patch gnome-panel_2.28.0.bb): gnome-panel: refresh idl-sysroot patch for 2.28.0 Oct 26 17:26:29 03Grégoire Gentil  07org.openembedded.dev * r7056d05cfe 10openembedded.git/ (3 files in 3 dirs): fbreader: add 0.10.7 Oct 26 17:26:36 03Grégoire Gentil  07org.openembedded.dev * re0fcc1bff9 10openembedded.git/recipes/gnome-mplayer/ (files/default.patch gnome-mplayer_0.9.8.bb): Oct 26 17:26:39 gnome-mplayer: tweak defaults Oct 26 17:26:41 * disable fullscreen animation, the intel and omapfb driver don't like it Oct 26 17:26:43 * reduce cache size so youtube movies don't take eons to start Oct 26 17:26:45 * enable single instance Oct 26 17:26:47 03Koen Kooi  07org.openembedded.dev * rc3f3c6f8a5 10openembedded.git/recipes/file-roller/file-roller_2.28.0.bb: file-roller: add unzip to RDEPENDS Oct 26 17:26:52 03Martin Jansa  07shr/merge * rfabf18120a 10openembedded.git/conf/distro/shr.conf: Oct 26 17:26:55 shr/merge: Fix toolchain with missing configs, please review and/or test which missing part was causing Illegal instructions/Segfaults Oct 26 17:26:58 Signed-off-by: Klaus Kurzmann Oct 26 17:31:24 eFfeM: you can set ALLOW_EMPTY_pkg to create a blank ipk that contains no files Oct 26 17:31:38 there are a few examples of those, e.g. more or less all of the task package Oct 26 17:33:04 pb__ thanks, will find them Oct 26 17:54:49 khem: any idea on this glibc compile failure? http://www.pastebin.ca/1644030 Oct 26 17:57:37 hmm Oct 26 17:57:48 still INIT: PANIC w/ gcc 4.1.2, binutils 2.19, and eglibc 2.9 Oct 26 17:57:59 I'm beginning to suspect eglibc Oct 26 17:58:11 will try glibc now Oct 26 18:03:47 did you manage to figure out where it was blowing up? Oct 26 18:04:27 if you can get an actual crash dump then it should be fairly straightforward to figure out who is to blame Oct 26 18:04:30 no, i took the easy way, hoping it would just vanish if i tweak build params :/ Oct 26 18:04:34 heh Oct 26 18:04:46 i always try to stay away from low level debugging Oct 26 18:04:52 but seems i have no chance this time Oct 26 18:05:00 if you send me a tarball of your rootfs then I can have a little play with it Oct 26 18:05:17 that'd be great. let me upload it to somewhere Oct 26 18:05:56 hmm, that's going to take until tomorrow. will ping you then Oct 26 18:06:01 righto Oct 26 18:06:05 thanks Oct 26 18:07:52 pb_: could you have a look at my glibc compile problem, please? Oct 26 18:09:25 that looks like either a broken gcc or some kind of CFLAGS problem Oct 26 18:09:39 is there anything unusual about your configuration? Oct 26 18:10:15 yes. I try eabi for armv4 with gcc 4.4 and recent binutils. Oct 26 18:10:41 what exactly have you changed in the conf files? Oct 26 18:11:32 "arm-angstrom-linux" doesn't look obviously like an eabi target, I would have expected to see something like "...-linuxgnueabi". Oct 26 18:11:54 though possibly angstrom does something special there, I don't know. Oct 26 18:11:59 http://pastebin.ca/1644051 Oct 26 18:14:00 maybe there are no CFLAGS defined for armv4.. Oct 26 18:14:08 that all looks reasonable enough, though I don't quite know how abi selection works on angstrom. Oct 26 18:14:25 you might try using DISTRO=minimal (or micro), or asking in #angstrom Oct 26 18:15:11 there are nearly no person with this kind of knowledge in #angstrom Oct 26 18:15:34 but talking with you already gave me some ideas. So thanks ;) Oct 26 18:15:42 righto :-) Oct 26 18:17:44 do I need to remove $BUILDDIR/tmp if I change DISTRO? Oct 26 18:33:37 http://www.qualcomm.com/news/releases/2009/091026_New_Qualcomm_Subsidiary.html Oct 26 18:46:50 thesing: i don't think so as things in the work dir carry both the machine and distro name Oct 26 18:52:30 03Koen Kooi  07org.openembedded.dev * r0bee6a2ed8 10openembedded.git/recipes/linux/ (9 files in 3 dirs): linux-omap-pm 2.6.29: sync touchbook patches with linux-omap kernel Oct 26 19:03:34 who's the admin for OE git server? florian? mickey? Oct 26 19:03:57 cbrake, Laibsch, mickeyl Oct 26 19:03:59 in that order Oct 26 19:04:07 Crofton|work also i think Oct 26 19:05:40 mickeyl: that reminds me about replacing my lost commit key. will speak to cbrake later Oct 26 19:06:14 mickeyl: thanks. how much RAM that box has and is it 32 or 64 bit OS? Oct 26 19:06:15 lrg: feel free to send me your new one Oct 26 19:06:28 denix: IIRC it's a virtual box Oct 26 19:06:50 mickeyl: ok, will do -forgot to mention this in Berlin Oct 26 19:06:55 np Oct 26 19:07:02 * Crofton just runs dns Oct 26 19:07:32 I'm fighting with "OOM" issues on my git server, which is virtual as well... Oct 26 19:08:34 denix: running apache ? XorA|gone fixed some VM OOM issues by restarting apache every day Oct 26 19:09:36 denix: the machine shows 1G RAM in top Oct 26 19:09:43 and yes, I think it's virtual Oct 26 19:09:52 virtual machine Oct 26 19:10:01 hmm, I can try restarting apache... but it clearly shows git-repack takes all the memory on large repos and dies... Oct 26 19:10:22 are you using packaged software? Oct 26 19:10:27 1GB here as well Oct 26 19:10:35 CentOS 64bit Oct 26 19:10:48 Could be a 64bit issue Oct 26 19:11:23 played with packedGitLimit/packedGitWindowSize w/o much success Oct 26 19:11:30 But it looks like the hardware we have is 64 bit as well Oct 26 19:12:10 do you see x86_64 in uname -a? Oct 26 19:12:23 /proc/cpuinfo is http://paste.debian.net/50030/ Oct 26 19:12:37 I'm not sure if this machine is running in 32bit or 64bit mode Oct 26 19:13:03 what does "file /usr/bin/git" say? Oct 26 19:13:34 ELF 64-bit LSB executable, AMD x86-64 Oct 26 19:13:49 right, so you are indeed in 64-bit mode Oct 26 19:14:06 32 bit here Oct 26 19:15:08 from git manual on packedGitLimit: Default is 256 MiB on 32 bit platforms and 8 GiB on 64 bit platforms Oct 26 19:15:59 I only have 1GB and it's burstable :( Oct 26 19:16:25 03Marcin Juszkiewicz  07org.openembedded.dev * r764574e16f 10openembedded.git/recipes/xorg-xserver/xserver-kdrive-nomadik_1.3.0.0.bb: xserver-kdrive-nomadik: mark as nhk-15 only Oct 26 19:16:27 tried adjusting packedGitLimit in the config file, but it still runs out of memory on large repos Oct 26 19:16:37 ka6sox put more RAM in the machine we are on recently .... Oct 26 19:16:56 denix: you could try installing a 32-bit git binary and see if your situation improves Oct 26 19:17:28 pb_: ah, interesting idea! haven't tried that yet. Oct 26 19:18:19 I don't know how easy centos makes that, but in principle there is no reason it can't be done. Oct 26 19:18:39 we used to run a 32-bit mysqld on an otherwise 64-bit system for fairly similar reasons Oct 26 19:19:09 03Marcin Juszkiewicz  07org.openembedded.dev * r6df5a87a70 10openembedded.git/recipes/sgalib/ (4 files in 3 dirs): sgalib: NHK-15 library for userspace access to graphics acceleration Oct 26 19:22:02 insanity Oct 26 19:22:11 the gpmc controller is on crack! Oct 26 19:29:01 re Oct 26 19:31:50 hi Oct 26 19:38:28 re Oct 26 19:51:54 denix, I can make the gpmc controller bring nOE and nWe low at the same time Oct 26 20:07:13 03Michael 'Mickey' Lauer  07org.openembedded.dev * ra782b9cbac 10openembedded.git/recipes/freesmartphone/vala-dbus-binding-tool-native_git.bb: vala-dbus-binding-tool-native: remove redundant inherit autotools Oct 26 20:07:22 03Michael 'Mickey' Lauer  07org.openembedded.dev * r19f1db57ac 10openembedded.git/recipes/freesmartphone/fso-monitord_git.bb: fso-monitord: remove dependency on frameworkd Oct 26 20:07:30 hi, can anyone recommend a simple photo slideshow app that works well in Angstrom or Poky? Oct 26 20:11:43 for picture in *; do display "${picture}"; sleep 5; killall display; done Oct 26 20:11:44 :) Oct 26 20:16:06 :-) Oct 26 20:17:51 03Marcin Juszkiewicz  07org.openembedded.dev * rad6f7be448 10openembedded.git/ (conf/checksums.ini recipes/tzdata/tzdata_2009p.bb): tzdata: added 2009p Oct 26 20:31:46 hrw: about? Koen mentioned your a hard core user of EVDev with touchscreens. Any ideas on getting nice rightclick delay support going? Oct 26 20:32:14 DJWillis: I prefer one click Oct 26 20:33:24 hrw: so no right click at all? Yep, I just have a need to get right click emu into this setup on request. Been trying to find a nice clean setup to use. One click works great with evdev. Oct 26 20:34:35 DJWillis: after over year with Poky I forgot that TS devices have more then one mouse button Oct 26 20:34:47 hrw ;) Oct 26 20:34:53 Answers that one ;-) Oct 26 20:35:16 Right click emulation does feel like a massive hack ;-) Oct 26 20:37:14 kgilmer: x or fb? Oct 26 20:37:32 x denix Oct 26 20:39:01 ah, x should be easy - there are plenty of image viewers... Oct 26 20:41:18 ok, time to bye Oct 26 20:41:20 bye all Oct 26 20:41:28 hrw: bye Oct 26 20:41:32 bye hrw Oct 26 20:54:37 denix, any suggestions? Oct 26 21:09:19 How to fix a patch file error in gpe-mixer Oct 26 22:18:21 bedtime here, cya all tomorrow & stay well ! Oct 26 22:25:24 bye all Oct 26 23:03:15 Do opkg maintainer(s) hang out here? Oct 26 23:03:33 Anyone know where, if not here? Oct 26 23:06:31 there are opkg maintainers? Oct 26 23:06:34 :) Oct 26 23:07:19 oh Oct 26 23:07:24 like that eh... Oct 26 23:11:35 while (*raw) { Oct 26 23:11:35 strncat( temp, raw++, 1); Oct 26 23:11:35 } Oct 26 23:11:49 he used strncat.... it must be safe Oct 26 23:12:36 waaay too many instances of this sort of thing: strncat(temp, provstr, strlen(provstr)); Oct 26 23:15:51 yeah, opkg's fondness for strncat() and strncpy() is a little bit weird. Oct 26 23:16:57 i've convinced my boss that opkg needs some of my time :) Oct 27 01:18:19 i am trying to configure mux in the kernel Oct 27 01:18:40 is there anything else i need to configure apart from mux.c and mux.h ? Oct 27 01:30:30 is kernel pin muxing turned off by default? for overo? **** ENDING LOGGING AT Tue Oct 27 02:59:56 2009