**** BEGIN LOGGING AT Wed Feb 21 02:59:57 2007 Feb 21 03:57:30 03bzhou * r5565 10optware/trunk/make/readline.mk: readline: newer config.sub Feb 21 04:10:18 03steveb * r5566 10optware/trunk/make/ftpd-topfield.mk: ftpd-topfield: Updated to 0.7.0 Feb 21 04:26:31 03dyoung * r5567 10optware/trunk/ (5 files in 2 dirs): Feb 21 04:26:31 Make CROSSTOOL_SCRIPT and CROSSTOOL_VERSION overridable in the new Feb 21 04:26:31 platform structure. Feb 21 04:39:57 03dyoung * r5568 10optware/trunk/ (3 files in 2 dirs): make crosstool-native variables overridable in the new platform structure. Feb 21 05:39:34 03bzhou * r5569 10optware/trunk/make/py-lxml.mk: py-lxml: 1.1.2 -> 1.2 Feb 21 08:29:10 gah Feb 21 08:29:13 configure: error: Feb 21 08:29:13 *** You must have either have gettext support in your C library, or use the Feb 21 08:29:13 *** GNU gettext library. (http://www.gnu.org/software/gettext/gettext.html Feb 21 08:29:30 I *have* the gettext library staged! Feb 21 08:46:40 bleh Feb 21 08:46:41 checking host system type... Invalid configuration `powerpc-linux-uclibc': system `uclibc' not recognized Feb 21 08:46:44 configure: error: /bin/bash ./support/config.sub powerpc-linux-uclibc failed Feb 21 08:51:50 ok, anyone got any clues on how to fix those two compile errors? Feb 21 09:16:49 mwester: ping? Feb 21 09:28:41 Anyone? Feb 21 09:52:32 03blaster8 * r700 10kernel/trunk/patches/2.6.21/ (16 files in 2 dirs): First commit broken 2.6.21-rc1 build (willfix) Feb 21 10:09:54 03blaster8 * r701 10kernel/trunk/patches/2.6.21/ (4 files in 2 dirs): Patchrinse NAS-100D ARTOP fix again Feb 21 10:31:11 03blaster8 * r702 10kernel/trunk/patches/2.6.20/ (4 files): Update to 2.6.20.1 Feb 21 10:31:11 03blaster8 * r703 10kernel/trunk/ (README STATUS patches/2.6.21/series): Update documentation, fix 2.6.21-rc1 series Feb 21 12:28:32 03oleo * r5570 10optware/trunk/platforms/ (packages-ddwrt.mk packages-wl500g.mk): platforms: add firmware-oleg to ddwrt and wl500g Feb 21 12:33:43 eno I suggest only one .mk per platform. eg platform/wl500g.mk Feb 21 13:38:59 NAiL: ping Feb 21 14:09:35 osas: pong Feb 21 14:24:12 03mwester * r704 10kernel/trunk/patches/2.6.20/defconfig: Feb 21 14:24:12 defconfig - demoted netconsole to a module, not useful in-kernel unless Feb 21 14:24:12 boot cmdline can be changed by users. Feb 21 14:26:38 NAiL: what is the issue with asterisk on ts101? Feb 21 14:26:44 and which version? Feb 21 14:29:55 The problem was that the makefile assumed that if the target was uclibc, it was either mips or arm Feb 21 14:30:08 the version is not 1.4 ;) Feb 21 14:30:12 I fixed it though Feb 21 14:30:41 there's now a mess of ifeq's in that makefile Feb 21 14:33:41 I see Feb 21 14:33:42 ok Feb 21 14:33:55 eno: ping Feb 21 14:34:22 NAiL: are you going to fix the 1.4? Feb 21 14:38:19 asterisk14 doesn't seem to do the same stuff Feb 21 14:38:58 mwester: good call. I'm trying to lean up my image. Feb 21 14:41:53 Oh ... yeah ... Feb 21 14:42:04 to many things in my head ... :p Feb 21 14:42:22 NAiL: lemme know if you have issues with it Feb 21 14:43:17 np Feb 21 14:43:39 I'm probably getting this one anyway: Feb 21 14:43:40 checking host system type... Invalid configuration `powerpc-linux-uclibc': system `uclibc' not recognized Feb 21 14:44:35 for which package? Feb 21 14:45:08 I'm getting that on a *long* list of packages Feb 21 14:45:12 and I can't figure out why Feb 21 14:46:48 check with eno ... Feb 21 14:48:02 eno: ping? Got any ideas? Feb 21 14:48:47 I think it's to early for him :-) Feb 21 14:49:10 give him another 2 hours ... Feb 21 15:08:57 NAiL: see readline.mk last change Feb 21 15:12:48 cp $(SOURCE_DIR)/common/config.sub $(READLINE_BUILD_DIR)/support/ Feb 21 15:12:57 That needs to be done on all the builds that fail for me? Feb 21 15:13:58 There's a lot of them Feb 21 15:16:58 or set GNU_TARGET_NAME to powerpc-linux Feb 21 15:17:13 in toolchain-ts101.mk Feb 21 15:19:49 yeah, but that causes others to break again Feb 21 15:21:27 i think ideal GNU_TARGET_NAME should be a triplet like ds101g powerpc-xxx-linux Feb 21 15:25:44 eno: if the .svn files are excluded from asterisk dir, it breaks the build (there's no way to compute ASTERISKVERSION) Feb 21 15:40:49 duh, sorry osas, please put it back Feb 21 15:41:04 bbl Feb 21 15:53:03 k Feb 21 16:00:22 03osas * r5571 10optware/trunk/make/asterisk14.mk: asterisk14: r55196 -> r55870 we need the .svn files in order to generate proper asterisk version Feb 21 16:39:18 checking host system type... Invalid configuration `powerpc-603e-linux-uclibc': machine `powerpc-603e-linux' not recognized Feb 21 16:46:54 it just does not understand "linux-uclibc" is os of the cpu-vendor-os triplet Feb 21 16:47:16 it guessed "uclibc" is the os Feb 21 16:48:54 what does powerpc-603e-linux or powerpc-linux break? for most optware packages that should work Feb 21 16:52:36 I'm starting a build with powerpc-linux now Feb 21 17:49:41 oleo: ping? Feb 21 17:50:40 eno: powerpc-linux looks like it's working better Feb 21 17:51:54 NAiL: great Feb 21 18:24:17 hello Feb 21 18:28:22 please change your nick. Feb 21 18:34:17 03nail * r5572 10optware/trunk/make/dnsmasq.mk: dnsmasq: Upstream upgrade, previous version is unfetchable Feb 21 18:35:38 * mwester thanks NAiL because the previous version was broken (BOOTP problems) :) Feb 21 18:35:51 ah Feb 21 18:35:57 it's been removed from the site Feb 21 18:36:07 Must have really been broken, then! Feb 21 18:36:11 earlier versions were accessible still Feb 21 18:36:24 but .36 and .37 were gone Feb 21 18:37:14 The last few versions added tftp support -- which now means that dnsmasq is sufficient in itself to do DHCP, DNS, and handle PXEBOOT. Sweet! Feb 21 18:37:34 03nail * r5573 10optware/trunk/platforms/toolchain-ts101.mk: toolchain-ts101: Change TARGET_OS to reduce breakage Feb 21 18:37:35 cool Feb 21 18:39:03 Yep. I can pxeboot any laptop or system here, and from a menu install Linux, run disk diagnostics, boot partition magic, use Ghost for backups (and of course, the backend servers are all NSLU2s. Feb 21 18:39:10 ) Feb 21 18:40:22 nice Feb 21 18:40:49 nice indeed Feb 21 18:41:18 btw - someone tried to do nfs rootfs over wifi? Feb 21 18:42:27 Isn't that like a high-wire act without a safety net? Even if you could make it work, is it really a good idea? Intriquing idea, though. Feb 21 18:43:29 You'd need to load the module from the iinitramfs or jffs I guess, then once the link was up and authenticated, continue the boot. Feb 21 18:43:51 too bad that progear webpad has to boot from internal hdd ;( Feb 21 18:57:20 hrw: I even swapped over nfs over wifi =) Feb 21 19:22:01 ok, next buildrelated issue: Feb 21 19:22:02 checking for iconv_open in -liconv... no Feb 21 19:22:02 configure: error: *** No iconv() implementation found in C library or libiconv Feb 21 19:22:27 I've built libiconv, so it should pick it up Feb 21 19:22:44 This is the same problem as I have with gettext, I assume Feb 21 19:22:49 eno: ping? Any idea? :) Feb 21 19:23:28 oh, and : Feb 21 19:23:28 In file included from charset.c:29: Feb 21 19:23:28 ./convert_unicode.h:16:19: iconv.h: No such file or directory Feb 21 20:19:06 03oleo * r5574 10optware/trunk/make/firmware-oleg.mk: firmware-oleg: fix mv instructions Feb 21 20:19:54 eno: pong Feb 21 20:20:28 oleo: got any idea on my iconv-issue? Feb 21 20:21:13 what about it? wchar problems? Feb 21 20:21:30 it doesn't pick up the staged libiconv Feb 21 20:22:27 I had the same problem with gettext earlier, even though I had gettext staged Feb 21 20:23:39 As I remember with uclibc I've had do do the patch for this. Feb 21 20:24:47 210-gettext-awareness.patch in sources/buildroot/uClibc-patches Feb 21 20:25:03 ok Feb 21 20:25:09 anything similar for iconv? Feb 21 20:25:36 this will solve your problems. Feb 21 20:25:58 Why are you using .27 instead of .28 Feb 21 20:26:40 because the target I'm building for is .27 Feb 21 20:27:54 But hey. Maybe you are not aware that all this buildroot thing runs out of core. Feb 21 20:28:05 huh? Feb 21 20:28:35 This means, that you can have system library .27 and still have the whole optware with .28 Feb 21 20:28:54 .27 and .28 is binary incompatible Feb 21 20:29:25 stuff keeps segfaulting if I build with .28 Feb 21 20:29:26 For example. oleg firmware has 0.9.19 but buildroot runs with 0.9.28 in /opt/lib Feb 21 20:29:39 ah Feb 21 20:30:04 maybe you have some LD_LIBRARY_PATH set? This should not happen. Feb 21 20:31:51 I created buildroot toolchain just for this purpose. For example: Kfurge created custom firmware for WL700g based on glibc system library. But still uses uClibc packages from oleg. Feb 21 20:41:33 Ok, trying with recent buildroot/uclibc Feb 21 20:48:16 If this is any similar to Dreambox, many packages will run out of the box with your package feed. Feb 21 20:58:59 It fails building a .28-toolchain Feb 21 20:58:59 adjtimex.c:14: error: '__adjtimex' aliased to undefined symbol 'adjtimex' Feb 21 20:58:59 adjtimex.c:15: error: 'ntp_adjtime' aliased to undefined symbol 'adjtimex' Feb 21 21:03:34 gah Feb 21 21:04:15 oleo: i've sent http://tech.groups.yahoo.com/group/nslu2-developers/message/1590 regarding wchar Feb 21 21:14:12 cd builds/libuclibc++; make menuconfig; See Wide Character Support (Not Currently Supported) under String and I/O Stream Support Feb 21 21:14:14 03bzhou * r5575 10optware/trunk/make/getmail.mk: getmail: 4.7.1 -> 4.7.2 Feb 21 21:15:24 fur such problems I suggest to use TARGET_GXX intead of TARGET_CXX Feb 21 21:15:45 I've done that in amule and rtorrent packages Feb 21 21:24:48 thx oleo, will try that Feb 21 21:45:28 oleo: Could you help me setting up a ppc-uclibc toolchain? Every time I try, it fails with the adjtimex-stuff above. Feb 21 21:52:46 hm. How should I start? Feb 21 22:12:53 No idea. I'm stumped. Feb 21 22:13:09 I've tried so many combinations to actually get aworking toolchain that I'm starting to give up Feb 21 22:13:27 What I would like to see is your config Feb 21 22:13:43 which config? Feb 21 22:15:50 ound config.powerpc in your distro Feb 21 22:16:37 ound? Feb 21 22:17:44 s/ound/found Feb 21 22:18:37 there's no config.powerpc Feb 21 22:21:48 there is sample config in your buildroot.tar.gz Feb 21 22:22:40 Still building powerpc versioin. Looks good. Feb 21 22:25:21 what kernel version are you using? Feb 21 22:25:58 2.6.12.3 Feb 21 22:27:00 I am now building for 2.4.32 Feb 21 22:29:01 Are 2.6.12.3 kernel headers any different than the stock 2.6.12.3 one? Feb 21 22:29:37 no Feb 21 22:36:36 03bzhou * r5576 10optware/trunk/ (4 files in 3 dirs): id3lib: patched and promoted for uclibc Feb 21 22:39:47 03bzhou * r5577 10optware/trunk/make/mediatomb.mk: mediatomb: enable-id3lib for uclibc Feb 21 22:39:55 hi Feb 21 22:41:41 morning Feb 21 22:43:53 NAiL: what timezone are you in? it's midnight here =) Feb 21 22:44:49 CEST Feb 21 22:44:56 ~ugt Feb 21 22:44:58 well, ugt is Universial 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 Feb 21 22:45:03 ;) Feb 21 22:45:53 NAiL: right, Feb 21 22:46:01 ~hail ugt Feb 21 22:46:03 * jbot bows down to ugt and chants, "I'M NOT WORTHY!!" Feb 21 22:46:04 =) Feb 21 22:52:13 oleo: I've build a toolchain with a recent buildroot manually. I'll try to get it work with optware again Feb 21 22:55:42 Ok. My build stopped at dl-elf.c due to undefined asm/mman.h: No such file or directory. Feb 21 22:56:02 Will try again with 2.6.12 headers Feb 21 23:01:02 hehe. I've als got error: '__adjtimex' aliased to undefined symbol 'adjtimex' with 2.6.12. Maybe another headers could solve this Feb 21 23:07:34 * CoreDump|home continues hacking on the samba wizzar Feb 21 23:07:36 d Feb 21 23:24:47 anybody tried soundcard on unslung? Feb 21 23:35:44 NSLU2-IRC: please change your nick Feb 21 23:37:01 sorry, OK now Feb 21 23:37:10 sure :-) Feb 21 23:37:26 well eny clues about sound on unslung? Feb 21 23:37:32 have you checked the wiki? Feb 21 23:37:42 yes all the way Feb 21 23:38:05 also, have you verified that your usb sound card is working ok under other linux distributions? Feb 21 23:38:07 osas: toolchain finished OK Feb 21 23:38:16 nice Feb 21 23:38:27 good to hear :-) Feb 21 23:38:28 dunno it it works as planned yet though ;-) Feb 21 23:38:29 i have sent a message on list with dmesg output Feb 21 23:38:45 simfun: I saw that Feb 21 23:39:03 I don't use a sound on my slug ... Feb 21 23:39:40 simfun: one completely unrelated tip: When starting a new thread, do not reply to other messages. Threading mail clients will then hide your mail in an unrelated thread (which people easily can ignore) Feb 21 23:39:47 have you loaded the driver? Feb 21 23:40:03 I see the following in your logs: drivers/usb/spca5xx.c: Devices not found !! Feb 21 23:40:26 well spca is not related Feb 21 23:40:36 it's about a webcam Feb 21 23:40:51 thanks NAIL Feb 21 23:41:32 oh ... Feb 21 23:42:48 well? Feb 22 00:00:05 i have to go Feb 22 00:00:30 if you know anything please post it on mailing list Feb 22 00:37:19 http://rafb.net/p/IUntlt86.html Feb 22 00:43:10 neat Feb 22 02:18:27 oleo: ping? Feb 22 02:20:50 oleo: My ts101 modifications to buildroot.mk is quite extensive. I'm considering changing it to pull patches and configs from sources/buildroot/$(OPTWARE_TARGET)/ instead of sources/buildroot/ (And the same with the uClibc stuff, of course) Feb 22 02:20:54 thoughts? Feb 22 02:21:26 (It'd reduce the "ifeq"s and sed-magic in the .mk, making it far easier to read) Feb 22 02:24:21 * NAiL is getting close to a good toolchain Feb 22 02:29:31 good enough to build amule, actually Feb 22 02:31:30 03nail * r5578 10optware/trunk/make/asterisk.mk: asterisk.mk: uclibc-builds are assumed to be either mips or arm. Bzzzt, wrong Feb 22 02:47:04 morning jacques Feb 22 02:48:41 howdy NAiL Feb 22 02:50:42 every time you join I think of jacques from "Finding Nemo" **** ENDING LOGGING AT Thu Feb 22 02:59:58 2007