**** BEGIN LOGGING AT Sat Oct 16 02:59:58 2010 Oct 16 04:08:20 I'm seeing: "WARNING: kmod-leds-gpio is not available in the kernel config" - what am I missing? Oct 16 06:07:18 cshore: ping? Oct 16 06:17:13 flyn * r23471 /packages/net/krb5/ (Makefile patches/ patches/001-krb5kdc-dir-to-etc.patch): Return 001-krb5kdc-dir-to-etc.patch; move database from /var to /etc for persistence Oct 16 06:25:46 nbd: ping? Oct 16 08:53:32 hey, my perl signal numbering patch (from 12 october) is languishing on openwrt-devel ml Oct 16 08:59:39 I feel your pain. :-) Oct 16 09:15:21 build #13 of ar71xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar71xx/builds/13 Oct 16 09:25:29 juhosg * r23472 /packages/lang/perl/ (Makefile files/config.sh-mips.in): (log message trimmed) Oct 16 09:25:29 fix perl signal numbering for mips Oct 16 09:25:29 This is a followup to my 16 month old patch for perl signal numbering for mipsel: Oct 16 09:25:29 https://dev.openwrt.org/changeset/16539 Oct 16 09:25:29 This time for mips. Tested on atheros (accton mr3201a). Also Oct 16 09:25:29 compared favorably to linux-2.6.git's arch/mips/include/asm/signals.h Oct 16 09:25:29 The most obvious benefit is that forked children get reaped Oct 16 09:33:09 danke! Oct 16 09:39:15 aha! Oct 16 09:39:45 * russell-- was wondering why i was seeing out-of-space errors when installing packages with opkg Oct 16 09:40:02 turns out it was a stale opkg.conf file of mine Oct 16 09:40:51 build #10 of ifxmips is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ifxmips/builds/10 Oct 16 09:42:23 option overlay_root Oct 16 10:13:02 build #10 of iop32x is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/iop32x/builds/10 Oct 16 10:45:56 build #7 of ubicom32 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/7 Oct 16 11:26:05 tripolar * r23473 /packages/net/ (4 files in 4 dirs): add maintainer to all of my packages Oct 16 11:38:15 philp64: pong Oct 16 11:38:37 sorry philipp64|laptop pong Oct 16 11:56:32 hi Oct 16 11:56:43 what for is this needed in packages? - PKG_INSTALL:=1 Oct 16 12:02:35 it calls make install with PKG_INSTALL_DIR as DESTDIR. Oct 16 12:13:25 thanks Oct 16 14:00:44 acinonyx * r23474 /trunk/package/mountd/ (3 files in 2 dirs): [package] mountd: Rename uci_add_history to uci_add_delta (#8084) Oct 16 14:10:58 mb * r23475 /packages/Xorg/lib/xine-lib/Makefile: xine-lib: Fix compile. Configure/build may not include headers from the host. Oct 16 14:20:28 build #7 of xburst is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/xburst/builds/7 Oct 16 14:54:01 mb * r23476 /packages/Xorg/xserver/xglamo/Makefile: xglamo: Depend on s3c24xx Oct 16 17:11:43 mb * r23477 /packages/utils/cdparanoia/Makefile: cdparanoia: Internal deps are broken. Force -j1 Oct 16 17:18:29 florian * r23478 /packages/libs/vips/ (. Makefile patches/ patches/001-embedded_build.patch): [package] add vips library (#7866) Oct 16 17:18:33 cshore: reping Oct 16 17:18:41 ping Oct 16 17:18:44 pong I mean Oct 16 17:22:51 I sent out several trivial patches... hoping some kind soul will get them committed. anything I need to do to increase my chances? Oct 16 17:24:24 also, I'm seeing: WARNING: kmod-leds-geos is not available in the kernel config Oct 16 17:25:07 I know where the message comes from, but I can't tell why it's being generated (the test is a little hairy). Oct 16 17:27:13 cshore: btw, what are the plans for the luci voice and devinfo diag mods? Oct 16 17:28:16 xMff: I'm planning on doing schedule scans which the luci frontend access rather than on-demands scans, because they take too long for the web if Oct 16 17:28:35 ok so further development is planned Oct 16 17:28:46 I currently purge some bitrot and wanted to ask :) Oct 16 17:30:18 xMff: no problem getting rid of it for now...it won't be much the same and isn't working right now Oct 16 17:31:55 xMff: but only in LuCI Oct 16 17:32:05 yes of course Oct 16 18:03:47 xMff: howdy. Oct 16 18:03:56 got a quick minute? Oct 16 18:04:44 sure Oct 16 18:06:20 I'm trying to add LEDs to the net5501 and Geos platforms. figured out why it wouldn't build for the net5501: they don't exist in .32. I'm building .35.7 for Geos, but they still won't build. just see the message above. Oct 16 18:07:41 the corresponding symbol is missing in the kernel config, it should at least appear as # blah is not set Oct 16 18:08:28 what's odd is that my .config contains: "CONFIG_DEFAULT_kmod-leds-geos=y" and "CONFIG_PACKAGE_kmod-"leds-geos=y". Oct 16 18:08:37 ah, ok. Oct 16 18:09:13 is there a project build farm to test my kernel version-bump patch? make sure no platforms break? Oct 16 18:10:00 yes. however if a kernel is bumped to a new micro version, all targets using it should be bumped Oct 16 18:11:41 the only thing I didn't test for was other processor families (Xscale, ARM, MIPS, PPC) having new specific flags. Oct 16 18:12:04 mb * r23479 /packages/net/mdnsresponder/Makefile: mdnsreponder: Internal deps are broken. Force -j1 Oct 16 18:12:08 the generic .35.7 and x86-specific flags were handled. Oct 16 18:12:49 but what didn't bump was the LINUX_VERSION:= in any of the /Makefile's. Oct 16 18:13:22 so you should be able to commit it without it breaking anything, since it doesn't get enabled without that last change... Oct 16 18:13:40 "if a kernel is bumped to a new micro version, all targets using it should be bumped" Oct 16 18:13:58 there is no point in keeping multiple 2.6.25.x around Oct 16 18:14:01 *35 Oct 16 18:15:40 build #8 of ps3 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/8 Oct 16 18:16:11 ok, so which version do you want it bumped to? Oct 16 18:16:25 35.7 ? Oct 16 18:16:43 actually *I* don't want any bumps Oct 16 18:16:54 but if you do so, do it for all users of 2.6.35.x Oct 16 18:18:16 ok, not following. I didn't think anyone was using it. the default version right now is 2.6.32, right? Oct 16 18:19:31 $ grep LINUX_VERSION target/linux/*/Makefile | grep 35 Oct 16 18:19:32 target/linux/au1000/Makefile:LINUX_VERSION:=2.6.35.7 Oct 16 18:19:32 target/linux/brcm63xx/Makefile:LINUX_VERSION:=2.6.35.7 Oct 16 18:19:32 target/linux/kirkwood/Makefile:LINUX_VERSION:=2.6.35.7 Oct 16 18:19:32 target/linux/malta/Makefile:LINUX_VERSION:=2.6.35.7 Oct 16 18:19:32 target/linux/omap35xx/Makefile:LINUX_VERSION:=2.6.32.24 Oct 16 18:19:32 target/linux/xburst/Makefile:LINUX_VERSION:=2.6.35.7 Oct 16 18:21:02 anyway, just checked your version bump patch, don't see any version change Oct 16 18:21:15 you're probably talking about something different Oct 16 18:22:59 see my email in the last hour titled "[PATCH] Kernel version bump to 2.6.35.7". I was adding all of the CONFIG_xxx values that the current global definitions are missing. Oct 16 18:24:46 for me a version bump is updating the version Oct 16 18:24:53 misleading title Oct 16 18:25:04 the initial patch I sent out included a hunk for target/linux/x86/Makefile to change the LINUX_VERSION to 2.6.35.7, but was told to take it out. Oct 16 18:25:15 forget who said that. maybe Florian or Felix. Oct 16 18:26:18 forgot I had sent it out a couple of days ago as "[PATCH v4] Add missing flags for building 2.6.35 on x86" Oct 16 18:27:13 losing my mind... or working too hard. Oct 16 18:28:56 can someone commit the CFLAGS:= ... -pipe ... patch? Oct 16 18:30:21 philipp64|laptop: there's a difference between the two? Oct 16 18:30:36 between which two? Oct 16 18:30:50 [19:24:30] losing my mind... or working too hard. Oct 16 18:31:24 ah. well, maybe cause and effect. or just getting old. Oct 16 18:35:12 xMff: fail. I added "# CONFIG_LEDS_GEOS is not set" to target/linux/x86/geos/config-default but that didn't do it. Oct 16 18:36:03 but looking at include/kernel.mk, it looks for %=m %=y %=n only. Oct 16 18:36:05 btw I'm planning on being around my morning and not in the afternoon (since that's to be dedicated to hard core work as opposed to 'soft' work like email and so on) Oct 16 18:36:48 possibly some time in the evening as well, as I will be on relax time then Oct 16 18:45:48 mb * r23480 /trunk/include/ (host-build.mk package.mk): Support force-disabling per-package parallel build with {PKG,HOST}_BUILD_PARALLEL:=0 Oct 16 18:46:26 mb * r23481 /packages/ (5 files in 5 dirs): Use PKG_BUILD_PARALLEL:=0 to force-disable parallel build Oct 16 18:49:11 mb * r23482 /trunk/include/ (host-build.mk package.mk): Fix typo Oct 16 18:53:39 xMff: ok, so the "# CONFIG_LEDS_GEOS is not set" needs to go where? Oct 16 19:22:55 mb__: ping Oct 16 19:28:15 philipp64|laptop: pong Oct 16 19:29:14 *poke poke* :) Oct 16 19:33:44 oh, sorry. was making coffee. Oct 16 19:33:51 it's noonish here. Oct 16 19:34:30 anyway, saw the title of your commit... I know there are a lot of broken makefiles out there that won't parallel build. Oct 16 19:35:11 I usually find it easier to just fix the package's makefile, keep a patch locally, but submit the patch upstream and hope it appears in a subsequent release. Oct 16 19:35:15 philipp64|laptop: I'm currently trying to figure those out. I fixed some of them and marked hard-to-fix packages with PKG_BUILD_PARALLEL:=0 Oct 16 19:36:04 Feel free to fix any or all of those packages. Broken ones are marked now for easy grepping :) Oct 16 19:36:18 well, let's hope that before too long we can scrub all of the packages that use it to not use it on their subsequent (version) bumps Oct 16 19:36:27 how many are there, anyway? Oct 16 19:36:57 how many of what? Packages that break on -jX? Oct 16 19:38:00 also if you are looking to fix broken package the buildbot currently reports broken packages http://tksite.gotdns.org:8010/broken_packages/ Oct 16 19:38:34 Yep, I noticed. Oct 16 19:39:04 mb__: yes. Oct 16 19:39:06 My current goal is to get a complete build with all packages running with CONFIG_PKG_DEFAULT_PARALLEL enabled. Oct 16 19:40:43 My current build is around 16GB. I was told a complete build takes about 20GB. So I guess I'm about 75% done. Oct 16 19:41:14 But let's see. I'm not in a hurry Oct 16 19:44:51 good luck with that. I know that a lot of project owners don't like to accept outside Makefile fixes, because they themselves are sometimes uncomfortable changing something they don't fully understand. and many use ':' when they really need '|'. Oct 16 19:44:56 I would actually like to know why this fails on some arches: http://tksite.gotdns.org:8010/broken_packages/x86/gpxview/compile.txt Oct 16 19:45:04 Need config.log to figure that out Oct 16 19:46:33 philipp64|laptop: Yeah, lots of people don't know how to write makefiles properly. They use it as extended shellscript. That fails horribly with parallel build. Most of those packages I marked as broken on parallel build use such shellscript-makefiles. Oct 16 19:46:54 I already fixed those that seemed to be easy to fix (just a missing dependency or two) Oct 16 19:48:05 I think the biggest offender is the configure script that OpenSSL uses... written in Perl and having nothing to do with autoconf. Oct 16 19:48:21 However, I don't see that as a big issue, yet, because all of those packages (so far) are really small and compile within a few seconds. So there's not that much benefit of parallelizing them anyway Oct 16 19:49:14 philipp64|laptop: The worst configure script I ever saw is the python-sip crapomatic. It needs _serious_ patching and hitting to work in crosscompile environment. Oct 16 19:49:14 yeah... many packages take more time to run ./configure than to compile and link, unfortunately. Oct 16 19:49:29 mb__: not sure how I could grab that in a automated way at this point Oct 16 19:49:41 build #10 of rdc is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/rdc/builds/10 Oct 16 19:49:41 and most of that time is spent computing the same answers over and over. too bad they can't all share an ac_cv_* cache. Oct 16 19:49:59 thepeople2: Well, can you manually cherrypick it? I'd really be interested in it, as I'm the maintainer of that package. Oct 16 19:50:10 sorry, grab which? Oct 16 19:50:27 config.log of gpxview package Oct 16 19:50:50 It compiles fine for me on ARM. Oct 16 19:51:17 mb__: nope, the buildbots don't have enough space on them to retain builds at this point, maybe I can cherry pick it with the log prepare script Oct 16 19:52:04 thepeople2: Ok, no need to jump through loops. I can also setup a test environment with x86 and reproduce. Oct 16 19:52:49 I think Jo-Philipp wandered off. I saw his answer about the "... is not available in the kernel config" question, but either I misunderstood his answer or it was missing something additional. Oct 16 19:54:44 thepeople2: I'm also unsure why QT4 is listed as broken package. The compile.txt log seems fine. I don't see any errors. Oct 16 19:54:47 http://tksite.gotdns.org:8010/broken_packages/x86/qt4/ Oct 16 19:56:46 mb__: libmng.cpp:42:20: error: libmng.h: No such file or directory Oct 16 19:57:02 is that a race condition, or just missing the correct -I flags? Oct 16 19:57:42 oh, wait... wrong package. Oct 16 19:58:10 mb__: make[3]: *** [libmng.o] Error 1 which is part of the qt build I would image Oct 16 19:58:38 but qt just disables it if it fails Oct 16 19:59:11 mb__: actually it looks like there may be a bunch of things in there with errors Oct 16 20:00:52 thepeople2: no, these are just config check failures Oct 16 20:00:57 These are expected Oct 16 20:01:31 It uses a non-autotools config script that dumps the result to stdout, even if the checks fail. Oct 16 20:01:36 why is config spewing errors to stderr rather than sending it to a capture file? sigh. Oct 16 20:02:03 Well, I dunno. That's the way their config script works. Oct 16 20:02:11 mb__: well I need more conditions in my script to look for errors then Oct 16 20:02:35 Ah so you look for stderr==empty >= build OK Oct 16 20:02:44 nope Oct 16 20:03:43 mb__: grep '\[*\] Error [0-9]' logs/* -R |grep -v "ignored" Oct 16 20:04:22 Ah yeah. Ok that is going to give false positives on QT4 Oct 16 20:04:40 so I see :-) Oct 16 20:05:14 Maybe the configure script can be fixed. I dunno how hard this would be. Oct 16 20:10:56 mb * r23483 /packages/Xorg/app/pwrtray/Makefile: pwrtray: MIPS fixes Oct 16 20:15:45 build #10 of rb532 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/rb532/builds/10 Oct 16 20:18:25 well, I guess I'm not figuring out the ".. is not available in kernel config" issue today. I'll try again tomorrow. Oct 16 20:29:23 mb * r23484 /packages/lang/python-sip/ (Makefile files/sipconfig.macros.template): python-sip: Compile with -fpic Oct 16 20:34:34 thepeople2: The perl build also seems to be fine: http://tksite.gotdns.org:8010/broken_packages/brcm47xx/perl/compile.txt Oct 16 20:37:05 mb__: make[5]: *** [lib/Config.pod] Error 126 Oct 16 20:37:15 So it seems we're doing pretty well. There are not that many broken packages and some of these are already fixed in trunk. Oct 16 20:39:14 thepeople2: Yeah, that seems to be some "configure" test. Oct 16 20:39:43 See this command: ./miniperl -w -Ilib -MExporter -e '' || make minitest Oct 16 20:40:04 mb__: I guess I have more parsing to do to see if it failed Oct 16 20:40:24 Probably hard to do Oct 16 20:41:14 mb__: yea, but maybe it can be done better Oct 16 20:47:17 mb * r23485 /trunk/target/linux/omap24xx/Makefile: omap24xx: Update kernel Oct 16 20:58:27 thepeople2: The VIM build is also a false positive Oct 16 20:59:45 mb__: ok Oct 16 21:15:15 mb * r23486 /packages/utils/maemo-kexec/ (Makefile files/sbin/boot-maemo files/sbin/bootmaemo): maemo-kexec: Rename script to bootmaemo. Oct 16 21:19:55 mb * r23487 /packages/utils/gnokii/Makefile: gnokii: Fix whitespace Oct 16 21:39:24 mb * r23488 /packages/utils/gnokii/Makefile: gnokii: Parallel build is broken Oct 16 21:46:06 mb * r23489 /packages/utils/btrfs-progs/Makefile: Fixes #8086 Oct 16 21:53:03 mb * r23490 /packages/utils/btrfs-progs/Makefile: btrfs-progs: Enable parallel build Oct 16 22:17:46 build #10 of kirkwood is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/kirkwood/builds/10 Oct 16 22:30:55 build #10 of au1000 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/au1000/builds/10 Oct 16 22:53:14 zandbelt * r23491 /packages/net/asterisk-1.6.x/Makefile: [packages] asterisk-1.6.6: added app_alarmreceiver,app_setcallerid and app_verbose Oct 16 23:01:00 build #12 of brcm63xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/12 Oct 16 23:10:39 blarg Oct 16 23:10:59 thepeople: can you not just test on exit code? Oct 17 00:31:56 build #10 of orion is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/10 Oct 17 00:35:52 build #10 of ixp4xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ixp4xx/builds/10 Oct 17 00:58:02 mb * r23492 /packages/net/gmediaserver/Makefile: gmediaserver: Fix parallel build Oct 17 01:05:19 mb * r23493 /packages/net/gmediaserver/Makefile: gmediaserver links to uclibcxx and zlib. So add these deps. Oct 17 01:54:00 finally getting around to looking at the wgt634u problems Oct 17 01:55:07 never see the target/linux/brcm47xx/files-2.6.34/drivers/mtd/maps/bcm47xx-flash.c printk()'s i inserted Oct 17 01:55:15 * russell_ wonders why Oct 17 02:26:35 build #10 of pxcab is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/10 **** ENDING LOGGING AT Sun Oct 17 02:59:58 2010