**** BEGIN LOGGING AT Mon Nov 17 02:59:57 2008 Nov 17 07:45:08 juhosg * r13253 /trunk/tools/firmware-utils/src/mkmylofw.c: [tools] firmware_utils/mkmylofw: add support of named partitions Nov 17 09:19:45 juhosg * r13254 /trunk/target/linux/ (6 files in 6 dirs): [ar71xx] move MyLoader stuff into generic files, will be used on adm5120 Nov 17 09:20:51 <_trine> morning all Nov 17 09:21:15 <_trine> I have just downloaded a new svn trunk onto my laptop Nov 17 09:21:49 <_trine> if I use the defaults in make menuconfig it compiles ok Nov 17 09:23:23 <_trine> if I then change 1 thing which is going from the default broadcom device to ahteros 231x/5312 it won't build Nov 17 09:24:56 <_trine> the eror is:- /trunk/staging_dir/mips/stamp/.toolchanin_install] Error 2 Nov 17 09:25:07 <_trine> sorry for the typos Nov 17 09:41:56 <_trine> http://pastebin.com/m26f296de Nov 17 09:43:10 <_trine> this is V=99 and the only change I did to make menuconfig was to change the atheros .. this is on a new svn download from trunk Nov 17 09:47:41 juhosg * r13255 /trunk/target/linux/adm5120/files/drivers/mtd/myloader.c: [adm5120] update MyLoader partition parser, add support of named partitions Nov 17 09:55:30 _trine: does it do the same thing if you build atheros on a clean tree? Nov 17 09:55:50 <_trine> thats what I did Nov 17 09:56:07 you said you build for something else first Nov 17 09:56:53 <_trine> yes Nov 17 09:57:08 try without doing that "first" thing Nov 17 09:57:11 <_trine> let me tell you exactly what I did Nov 17 09:57:23 <_trine> ok Nov 17 09:57:38 <_trine> I will delete it all Nov 17 09:58:01 to determine if it is really specific to building atheros or if it has to do with the sequence Nov 17 09:58:24 <_trine> but the strange thing is it compiles ok on my main machine which has been working for ages now Nov 17 10:00:23 <_trine> ok I am downloading it all again Nov 17 10:00:27 <_trine> 1 minute Nov 17 10:01:02 <_trine> when I have it I will change to atheros right from the start Nov 17 10:01:09 <_trine> and try a compile Nov 17 10:01:24 <_trine> but I am quite sure that's what I did last time Nov 17 10:01:35 <_trine> anyway lets see Nov 17 10:04:27 <_trine> making now Nov 17 10:08:10 <_trine> russell--, it seems to have got further this time Nov 17 10:08:41 <_trine> the only difference was that the first time I tried it I compiled broadcom first Nov 17 10:08:55 <_trine> then it failed on atheros Nov 17 10:09:01 yeah, so perhaps something got touched during that that threw off the second build Nov 17 10:09:13 <_trine> looks like it Nov 17 10:09:49 <_trine> there was someone else yesterday with the same problem Nov 17 10:10:24 <_trine> I remember because I tried a build on mine and it was ok on my main machine Nov 17 10:10:59 <_trine> it's just that I'm rebuilding my laptop that I have noticed this Nov 17 10:11:23 script -c 'make V=99' typescript-1 Nov 17 10:11:38 <_trine> basically I just wanted the toolchain on my laptop Nov 17 10:12:05 etc, then diff'ing the typescripts might give you a clue where things go wrong Nov 17 10:12:19 <_trine> yes Nov 17 10:15:13 juhosg * r13256 /trunk/target/linux/ (9 files in 7 dirs): [adm5120] move MyLoader partition parser into generic files, will be used on ar71xx Nov 17 10:23:57 juhosg * r13257 /trunk/target/linux/ (4 files in 3 dirs): [ar71xx] enable Myloader partition parser Nov 17 10:47:06 jow * r13258 /trunk/package/base-files/files/lib/network/config.sh: Prevent "SIOCSIFHWADDR: Device or resource busy" when trying to change mac address Nov 17 11:45:53 <{Nico}> _trine: ping Nov 17 11:46:45 <_trine> hello Nov 17 11:46:51 _trine: huh? Nov 17 11:47:53 <{Nico}> _trine: you build problem when swithing targets is a known issue: wrong target cflags are kept in .config Nov 17 11:48:59 <{Nico}> _trine: so you have to remove .config when before switching targets with different arches, until we find a way to handle this Nov 17 11:49:21 <_trine> ah OK thanks for letting me know Nov 17 17:04:17 juhosg * r13260 /trunk/target/linux/ar71xx/ (config-2.6.26 config-2.6.27): [ar71xx] remove root cmdline option Nov 17 17:22:41 kaloz * r13261 /trunk/ (4 files in 4 dirs): update to 2.6.27.6 Nov 17 17:23:28 kaloz * r13262 /trunk/target/linux/generic-2.6/patches-2.6.27/208-rtl8110sb_fix.patch: resync patch Nov 17 17:24:34 kaloz * r13263 /trunk/target/linux/ppc40x/ (Makefile patches/001-kilauea_openwrt_flashmap.patch): upgrade ppc40x to 2.6.27 Nov 17 17:28:35 kaloz * r13264 /trunk/target/linux/ (9 files in 9 dirs): move some options to the generic 2.6.27 config Nov 17 17:42:40 juhosg * r13265 /trunk/target/linux/ar71xx/ (3 files in 3 dirs): [ar71xx] use MyLoader partition parser on the WP543 board Nov 17 18:04:44 kaloz * r13266 /trunk/target/linux/ppc44x/ (3 files in 2 dirs): upgrade ppc44x to 2.6.27 Nov 17 18:53:33 I'm trying to compile Kamikaze 8.09 and I'm getting a compliation error: Nov 17 18:53:37 ../libtool: eval: line 4350: syntax error near unexpected token `|' Nov 17 18:55:59 this is a vanilla SDK and I'm building it for RDC architecture Nov 17 19:29:07 cry0x: try trunk Nov 17 19:29:56 will do Nov 17 19:37:56 for the record, I love this SDK Nov 17 19:42:16 Support for AMIT devices... good stuff! Nov 17 19:43:47 kaloz * r13267 /trunk/package/madwifi/Config.in: MADWIFI_DEBUG should depend on DEVEL Nov 17 19:46:39 juhosg * r13268 /trunk/target/linux/atheros/config-2.6.26: [atheros] sync 2.6.26 config Nov 17 19:50:44 what bootloader is used by default in trunk? Nov 17 19:51:18 I'm using a modified redboot and I want to make sure it won't overwrite it Nov 17 19:52:09 florian * r13269 /trunk/tools/lzma/Makefile: Directly use the kent sourceforge mirrors, previous are in timeout Nov 17 19:52:27 <[florian]> cry0x: default bootloader depends on your target Nov 17 19:52:39 <[florian]> cry0x: most targets use u-boot or redboot, broadcom based device use CFE Nov 17 19:53:05 [florian], is there a way to determine which bootloader? I'm building for RDC Amit devices Nov 17 19:53:32 <[florian]> cry0x: ah, these are the RDC devices that I could not get my hands on yet Nov 17 19:53:40 <[florian]> cry0x: ask sn9 when he comes online Nov 17 19:53:45 will do :) Nov 17 19:54:03 These devices aren't so wonderful, by the way Nov 17 19:54:16 <[florian]> if I recall rights, those are not using a redboot loader but something pretty crappy Nov 17 19:54:17 at least they'll be running good software Nov 17 19:54:20 <[florian]> not at all Nov 17 19:54:30 <[florian]> they have two advantages Nov 17 19:54:40 <[florian]> easy pcb design, and x86 compatibility Nov 17 19:54:51 AMIT has a proprietary bootloader. I replaced it with a modified Redboot so I could mess around with stuff Nov 17 19:54:56 <[florian]> the SoC being fairly low-cost Nov 17 19:55:02 <[florian]> yeah, good idea :) Nov 17 19:55:11 it's very cheap. A lot of stuff is broken though Nov 17 19:55:19 cold solder joints on the wifi module are crap Nov 17 19:55:30 the USB as a 1v ripple Nov 17 19:55:41 <[florian]> lol Nov 17 19:55:51 one volt... I thought my gague was wrong Nov 17 19:56:27 <[florian]> usb is 5V or 0V Nov 17 19:56:58 anyway, I'm working on getting this onto a 3G amit device. I'll probably have to modify a lot so I guess here's my chance to give back to the community Nov 17 19:56:59 5v yeah Nov 17 19:57:09 Sometimes a multimeter is not the right tool to measure modern electronics.. Nov 17 19:57:29 xrg, what would you recommend to check the power on the usb port? Nov 17 19:57:46 this device has a lot of issues with USB 3G cards and I figured it was power related Nov 17 19:57:53 Osciloscope, always! Nov 17 19:58:09 You may find things you never knew they were there.. Nov 17 19:58:41 It's an osciloscope Nov 17 19:58:45 I have found, 2 times, a blown fuse that would appear as connected in the multimeter, but blown at the osciloscope. Nov 17 19:58:47 however, I'm not an EE Nov 17 19:59:20 of course, But when you dive into hardware, tools matter. Nov 17 19:59:35 true true Nov 17 19:59:41 I'm always looking to expand my knowledge Nov 17 19:59:54 You could notice spikes or ripples at the power (which would need to be 5V stable).. Nov 17 20:00:45 Sometimes, it is even necessary to use the 10x divider on the probe, so that you don't itroduce capacitance on the circuits you try to probe. Nov 17 20:17:02 [florian], do you want an AR3603GW? Nov 17 20:17:13 aka WMQ137 Nov 17 20:17:37 <[florian]> oh yes, if you do not use it Nov 17 20:17:48 <[florian]> do you want me to send you an ar525w in exchange ? Nov 17 20:18:45 right now my project is ar360w3g support, so I wouldn't use a 525, but thanks Nov 17 20:19:18 I have a board with some serial cables attached for debugging, I can remove them if you want Nov 17 20:20:07 <[florian]> soldering one is not a problem, but thanks Nov 17 20:47:20 florian: you bumped au1000 to 2.6.27 some days ago ? Nov 17 20:47:27 <[florian]> yes Nov 17 20:47:30 <[florian]> this weekend Nov 17 20:47:50 thanks for that. but i think you missed something. Nov 17 20:47:58 <[florian]> tell me Nov 17 20:48:00 i get oops since then. Nov 17 20:48:09 <[florian]> yes, on the gpio code, which is harmless Nov 17 20:48:14 yes. Nov 17 20:48:20 <[florian]> I am converting the board to using gpiolib Nov 17 20:48:47 <[florian]> is there anything else that you noticed ? Nov 17 20:48:59 and i looked at the ethernet code, you took my patch (phy, led) and some of the other things seem to be fallen out (link beat, ...). Nov 17 20:49:21 <[florian]> ah right, it still needs the original patch Nov 17 20:49:39 <[florian]> this ethernet driver needs a bit of love Nov 17 20:49:58 ;) Nov 17 20:50:43 i wondered when i noticed that openwrt has a completely different one than oe has. Nov 17 20:51:16 <[florian]> I used oe patches in the early beginning of the au1000 port, then decided not to use their non-standard code Nov 17 20:51:30 ah, ok. Nov 17 20:52:34 i have also a pathc for making i2c working via gpio, but it is not good enough for beeing submitted. you also tried this some time ago, why did you remove your version ? Nov 17 20:52:51 <[florian]> it appeared that no one was actually using i2c Nov 17 20:52:58 <[florian]> so I simply dropped the patch :) Nov 17 20:54:37 hmmm, ok. then i won't bother you with my version of that ;) Nov 17 20:55:29 <[florian]> no, if you use it, it should be included Nov 17 20:55:39 <[florian]> one user is enough for me to keep the patch there Nov 17 20:57:04 i'll look at your version, it's likely to be the better one i think, since i am not an experienced kernel hacker. Nov 17 20:58:12 <[florian]> in fact, there is a gpio i2c driver from juhosg in package/ which is probably even better Nov 17 20:59:09 i noticed it, but havn't tried it yet. Nov 17 21:08:12 florian * r13270 /packages/net/quagga-unstable/ (. Makefile files/ files/quagga files/quagga.init): Add the unstable version of quagga based on 0.99.11 Nov 17 21:09:42 trunk build still has the same compilation error with libtool: Nov 17 21:09:50 ../libtool: eval: line 4350: syntax error near unexpected token `|' Nov 17 21:10:51 during the compilation of opkg Nov 17 21:11:21 <[florian]> weird Nov 17 21:13:44 I'm not using any odd settings... I'll try with a default config Nov 17 21:18:37 here's the error: http://pastebin.ca/1259718 Nov 17 21:19:26 there's a syntax error with the double pipe | | Nov 17 21:25:31 <[florian]> strange Nov 17 21:26:27 yep. I figure that something is supposed to be between those Nov 17 21:47:15 I'll try building on a different box :( Nov 17 21:49:35 linux Nov 17 22:04:41 i thought somebody highlighted me, but i don't find the line ^^ Nov 17 22:04:42 florian: i tired to get the i2c driver in package/ selectable, but it seems that there might be another config quirk. i re-added GENERIC_GPIO to Kconfig, that brings me back the kernel's i2c gpio thing, but juhosg's driver doesn't show up anywhere :( Nov 17 22:11:42 i just selected another target (AR71xx) for testing, and the driver shows up where it should. re-selecting AU1x00 makes it disappear again. Nov 17 22:44:08 lars * r13271 /packages/Xorg/lib/libpciaccess/ (. Makefile): Add libpciaccess. Nov 17 23:00:31 lars * r13272 /packages/Xorg/lib/libpciaccess/Makefile: Install headers in DevInstall rather then install. Nov 17 23:22:40 jow * r13273 /trunk/package/mmc_over_gpio/ (Makefile files/mmc_over_gpio.config files/mmc_over_gpio.init): make mmc_over_gpio init uci-aware Nov 18 00:13:10 what the heck... trying to compile openwrt trunk and I'm still getting the libtool syntax error Nov 18 00:13:32 vanilla trunk, completely different computer now running a different OS (ubuntu instead of arch) Nov 18 00:14:21 http://pastebin.ca/1259859 Nov 18 00:28:17 lars * r13274 /packages/Xorg/driver/xf86-video-vesa/patches/ (. 010-dont-mess-up-include-paths.patch): Fix xf86-video-vesa compile. Nov 18 00:35:16 cry0x: "export_symbols_cmds="\$NM \$libobjs \$convenience | \$global_symbol_pipe | \$SED 's/.* //' | sort | uniq > \$export_symbols" Nov 18 00:35:45 global_symbol_pipe is a sed expression for me Nov 18 00:36:07 larsc, thanks, I'll try it Nov 18 00:36:13 take a look at your libtool and see if its empty Nov 18 00:36:45 cry0x: that wasn't a solution ;) its the cmd that causes your problem Nov 18 00:37:29 :O Nov 18 00:38:48 cry0x: open /home/aport/nxa/build/openwrt/trunk/build_dir/i386/opkg-4564/libtool and search for the definition of global_symbol_pipe Nov 18 00:39:10 will do. I'm on a new computer so I'm checking out trunk right now Nov 18 00:39:15 ok Nov 18 00:40:32 making :) Nov 18 00:42:49 global_symbol_pipe="sed -n -e 's/^.*[ ]\\([ABCDGIRSTW][ABCDGIRSTW]*\\)[ ][ ]*\\([_A-Za-z][_A-Za-z0-9]*\\)\$/\\1 \\2 \\2/p'" Nov 18 00:42:59 looks ok Nov 18 00:43:13 and build still fails? Nov 18 00:43:17 indeed Nov 18 00:43:57 sed --version ? Nov 18 00:44:00 this is for RDC Nov 18 00:44:20 Bartman007, 4.1.5 Nov 18 00:46:59 cry0x: there are two definitions of gloabl_symbol_pipe what does the second say Nov 18 00:50:17 one moment Nov 18 00:52:44 they're all the same Nov 18 00:53:25 hm, no idea then... Nov 18 00:53:45 when you search for 'sort | uniq' what do you find? Nov 18 00:54:41 the syntax error looks like the double pipes | | Nov 18 00:54:48 it seems like there should be something in between them Nov 18 00:55:16 yes Nov 18 00:55:39 on my system its gloabl_symbol_pipe Nov 18 00:55:49 how about that Nov 18 00:57:12 so, when you search for 'sort | uniq' in the libtool file, what do you find? Nov 18 00:59:14 export_symbols_cmds="\$NM \$libobjs \$convenience | \$global_symbol_pipe | \$SED 's/.* //' | sort | uniq > \$export_symbols" Nov 18 01:00:51 ohhh snap Nov 18 01:00:56 no we're good Nov 18 01:01:14 when I configured it for RDC, global_symbol_pipe is set to nothing Nov 18 01:01:28 for a different build, it has the value posted above Nov 18 01:02:48 that's odd Nov 18 01:04:23 well, ok try to find out, why its set to nothing. the file is generated during libtool package build. i have to go and get some sleep now. Nov 18 01:04:36 thanks Nov 18 01:04:44 good night Nov 18 01:50:37 <{Nico}> larsc, cry0x: i found the problem with opkg & libtool Nov 18 01:51:07 {Nico}: I'll test bb 1.12.2 shortly Nov 18 01:51:23 <{Nico}> Bartman007: hey, cool Nov 18 01:51:34 <{Nico}> it seems to work fine on uml Nov 18 01:51:52 <{Nico}> larsc: still around? Nov 18 01:51:57 currently testing hostapd 0.6.5, if it looks good I'll commit it to trunk **** ENDING LOGGING AT Tue Nov 18 02:59:56 2008