**** BEGIN LOGGING AT Wed Oct 06 02:59:57 2010 Oct 06 05:52:40 what's the command to blow away everything but the toolchain? Oct 06 05:54:42 make dirclean ?! Oct 06 06:07:52 build #4 of brcm47xx is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/brcm47xx/builds/4 Oct 06 06:07:53 build #4 of sibyte is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/sibyte/builds/4 Oct 06 06:07:55 build #4 of etrax is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/4 Oct 06 06:22:07 blogic: no, that kills the toolchain as well. oh, well. Oct 06 06:22:23 then make clean :D Oct 06 06:22:28 i always confuse the 2 Oct 06 06:25:22 ugh. i hate mornings. Oct 06 06:30:34 why ? Oct 06 06:30:36 i love them Oct 06 06:30:43 most productive part of the day :) Oct 06 06:35:21 blogic: do you know the show "how i met your mother"? Oct 06 06:36:05 blogic: "nothing good happens after 2am" is one of the most known quote from it. I think we could add "nothing good happens before 10am" as well Oct 06 06:36:51 i'd say nothing good happens between about 5am and 1pm :P Oct 06 06:40:22 ok, off to work. back in an hour or so.. Oct 06 07:17:48 hi. Am I the only one getting a OOM on mips malta boot for trunk build? Oct 06 08:08:52 fscking rain.. Oct 06 08:15:05 hello Oct 06 08:15:44 yes, it was just me getting the OOM during boot time. make clean && rm .config* did the trick Oct 06 08:22:46 in the buildroot there is the files folder to replace the files on the final image Oct 06 08:22:55 but is there anything to automatically aply patches? Oct 06 08:23:39 for the case where I just want to replace part of a file, and not be harcoding it entirely Oct 06 08:46:28 not really sure what you mean... there is a thing for applying patches to packages.. Oct 06 08:51:33 trunk/files/* Oct 06 08:51:53 when you type make, the image files are replaced by whatever you have on this dir Oct 06 08:52:01 trunk/patches/* Oct 06 08:52:23 when you type make, the image files are patched by whatever patches are on this dir Oct 06 08:52:33 this is what I was asking for Oct 06 08:53:07 ok. Oct 06 08:57:29 morning _trine Oct 06 09:41:31 hcg * r23269 /trunk/target/linux/generic/patches-2.6.36/930-crashlog.patch: kernel: crashlog uses bootmem to allocate memory which is being deprecated in favour of memblock. microblaze, sh, arm, sparc and powerpc only have memblock. Oct 06 09:58:25 I'm trying to compile the click modular router for openwrt and I'm getting the following error: (.text+0xc): undefined reference to `__libc_csu_fini'. The gcc version I'm using is 4.4.4. Oct 06 09:59:00 the line that is generating the error is the following: g++ -O2 -pipe -march=i486 -funit-at-a-time -MD -fno-stack-protector -static -L/home/rriggio/src/kamikaze/staging_dir/target-i386_uClibc-0.9.30.1/usr/lib -L/home/rriggio/src/kamikaze/staging_dir/target-i386_uClibc-0.9.30.1/lib -L/home/rriggio/src/kamikaze/staging_dir/toolchain-i386_gcc-4.4.4_uClibc-0.9.30.1/usr/lib -L/home/rriggio/src/kamikaze/staging_dir/toolchain-i386_gcc-4.4.4_uClibc-0.9.30.1 Oct 06 09:59:02 /lib -o click-mkmindriver -rdynamic click-mkmindriver.o ../lib/libclicktool_build.a Oct 06 09:59:47 the system is actually supposed to use g++ since the executable that is building is part of the package build process Oct 06 10:52:17 after adding files in files/, what needs to be run to add them to the image? Oct 06 10:52:31 make package/rootfs-prepare targer/install? Oct 06 10:52:44 *target/install Oct 06 13:11:55 jow * r23270 /trunk/package/dnsmasq/ (Makefile files/dhcp.conf): [package] dnsmasq: allow 127.0.0.1/8 upstream DNS responses by default, resolves issues with Netflix and other services Oct 06 13:15:33 jow * r23271 /branches/backfire/package/dnsmasq/ (Makefile files/dhcp.conf files/dnsmasq.init): [backfire] merge r22487, r22490, r22528, r22592 and r23270 Oct 06 13:25:29 jow * r23272 /trunk/package/dnsmasq/files/dnsmasq.init: [package] dnsmasq: remove a useless subshell invocation from the init script Oct 06 13:26:10 jow * r23273 /branches/backfire/package/dnsmasq/files/dnsmasq.init: [backfire] merge r23272 Oct 06 13:26:58 Hi. There's something weird happening with my owrt mips malta build (for qemu): the pcnet driver gets compiled in a weird way (using different kenel headers??). When I try to insmod it it complains about the module_layout symbol. Oct 06 14:12:03 build #4 of at91 is complete: Failure [failed shell_3] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/4 Oct 06 14:12:35 sorry, got logged off, did I miss the response to my pcnet32 driver problem? Oct 06 14:12:42 nope Oct 06 14:13:25 :) Oct 06 14:13:28 thanks Oct 06 14:17:24 heil * r23274 /packages/net/haproxy/ (Makefile files/haproxy.init): [package] haproxy: fix blocking reload of service Oct 06 14:19:23 It seems to be a CONFIG_MODVERSION issue. Oct 06 14:31:22 build #4 of ixp4xx is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/ixp4xx/builds/4 Oct 06 14:31:23 build #4 of pxcab is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/4 Oct 06 14:31:25 build #4 of ifxmips is complete: Exception [exception failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/ifxmips/builds/4 Oct 06 14:31:27 build #4 of iop32x is complete: Exception [exception shell_3 failed slave lost shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/iop32x/builds/4 Oct 06 14:54:23 Hi there. I think that Rooslan's gripe about reducing lua arithmetic from double to float is quite legitimate. 10% decrease in memory use does in no way justify the introduction of such a nasty error-prone feature. Are there any valid arguments against removing the precision-reduction? Oct 06 14:56:15 heil * r23275 /packages/net/haproxy/files/haproxy.init: [package] haproxy: ash dont support $(<..) so change to $(cat ..) Oct 06 14:57:33 epointsystem: the original goal was to increase the speed on non-fpu systems, the 10% memory decrease was a nice side effect Oct 06 14:59:23 epointsystem: Lua's internal number representation is double (or float) without lnum Oct 06 14:59:39 epointsystem: have you tried whether Lua with LNUM but double solves your issue? Oct 06 15:02:03 No, we haven't tried that. Oct 06 15:02:21 then do that please. I'd rather not remove lnum, that'd slow Lua down to a crawl Oct 06 15:03:02 So, how would that be done? Oct 06 15:03:14 moment... Oct 06 15:06:36 try adding -DLNUM_DOUBLE to the CFLAGS Oct 06 15:06:53 apparently we use pure defaults atm which would be float + int32 Oct 06 15:08:49 Ok, that'd actually make sense. Oct 06 15:09:56 Because conversion of int to double would turn LNUM into a transparent optimization, which would speed up things considerably on platforms without an fpu and slow them down a little bit on platforms with one. Oct 06 15:10:03 correct Oct 06 15:10:20 LNUM with float + int32, however, changes the execution of code. Oct 06 15:10:22 btw, we had issues in luci too with that if numbers approached max_int Oct 06 15:10:41 we worked that around by splitting it into 2x 16bit ints :P Oct 06 15:10:50 it was only in two places Oct 06 15:11:09 Well, we have solved our problems by converting parts to C, but it's ugly as hell. Oct 06 15:11:55 Calculation with unix time are to be expected on a router, aren't they? Oct 06 15:11:56 yes, can imagine that Oct 06 15:19:58 Testing... Oct 06 15:22:02 BTW, the problem is with all the subtle little losses of precision that you DON'T notice, because stuff seems to work correctly. Converting ints into floats in a transparent manner without any control on the part of the application programmer is extremely dangerous. Oct 06 15:22:52 yes I agree and I can iamgine that it wilmake problems for your use case Oct 06 15:23:05 but for simple web programming it has not been a big deal until now Oct 06 15:23:11 Okay, test completed. LNUM with double does solve our problem. Oct 06 15:23:19 great Oct 06 15:27:45 In that case, please change the corresponding makefile in OpenWrt trunk and backfire branch. If you want to, I can submit the corresponding patch. Oct 06 15:28:16 I'll need to check the other stuff linking against Lua Oct 06 15:28:32 Please do. Oct 06 15:28:58 ok, its on my ToDo, will commit the change this night Oct 06 15:29:04 or tomorrow most likely Oct 06 15:46:54 Thanks a lot! Oct 06 17:02:27 failsafe on TL-WA901ND, anybody who can confirm that it really works? Oct 06 18:01:53 ifdown wan is blocking when wan is a 3g-ppp Oct 06 18:31:57 [florian]: hopefully this patch will be palatable. :-) Oct 06 18:33:05 nunojpg: can you please paste the "ps" output when its blocking? Oct 06 19:04:54 xMff: I will later...sorry for the delay Oct 06 19:21:12 mb * r23276 /packages/utils/retutahvo/ (7 files in 3 dirs): Add retutahvo tool. Oct 06 19:36:12 mb * r23277 /packages/utils/retutahvo/Makefile: retutahvo: Fix deps Oct 06 20:13:50 hauke * r23278 /trunk/target/linux/brcm47xx/ (20 files in 3 dirs): brcm47xx: add patches for kernel 2.6.36 Oct 06 20:57:41 hm. well, i have no real understanding of how to create a custom image file (or a custom image file builder) for openwrt.. though i shall continue to hit my head against the code. Oct 06 21:15:54 ping RealOpty Oct 06 21:23:03 jow * r23279 /trunk/package/base-files/ (Makefile files/lib/network/config.sh): [package] base-files: add udhcpc vendorid option (#7744) Oct 06 21:23:51 jow * r23280 /branches/backfire/package/base-files/ (Makefile files/lib/network/config.sh): [backfire] merge r23279 Oct 06 21:45:57 jow_laptop: do we use "send dhcp-client-identifier = hardware;" in dhclient by default? Oct 06 21:47:15 we don't even use dhclient Oct 06 21:50:03 philipp64: dnsmasq Oct 06 21:50:26 wrong, udhcpc Oct 06 21:50:40 doh, I'm thinking server Oct 06 21:52:35 * cshore_ thinks he's behind on the rent on his current brain....spare memes anyone? Oct 06 22:15:11 Hi, I've a problem with spi_register_driver used with a spi lcd screen. Oct 06 22:15:58 When kernel starts the driver should be initialized, but the /dev/fb0 (frame buffer) is not created so I can't access to this peripheral. Oct 06 22:16:31 can someone help me to investigate? Now I pastebin the code block where I found the problem. Oct 06 22:17:51 code block is here : http://openwrt.pastebin.com/PM5VjkGb Oct 06 22:18:32 you left out the interesting bits Oct 06 22:18:56 there's nothing that can really fail in this code besides the wodefb_probe() function Oct 06 22:22:59 xMff I was thinking about probe... Oct 06 22:23:18 what does expect the system from wodefb_probe ? Now I paste also this Oct 06 22:24:49 xMff: here is http://openwrt.pastebin.com/YrbZQKnN Oct 06 22:32:49 xMff: same question then... do we send a client-id generated from : or no? Oct 06 22:33:12 don't think so Oct 06 22:33:26 could we add an option to do so? Oct 06 22:34:10 I've worked with a few broken DHCP servers that don't handle xx:xx:xx:xx:xx:xx nearly as well as "01:xx:xx:xx:xx:xx:xx" Oct 06 22:34:30 including IOS going back to 12.2.... Oct 06 22:35:14 option clientid ... in your dhcp iface spec Oct 06 22:36:39 jow * r23281 /trunk/tools/Makefile: [tools] reenable squashfs4 for ar71xx, it needs both old and new Oct 06 22:41:09 so you have to know your MAC address then? Oct 06 22:44:05 I think so, yes Oct 06 22:46:45 would need to read the udhcpc code to know what it sends by default if no option is given, never bothered Oct 06 22:48:30 xMff: Do you know the syntax for using 2 ips (dnsmasq) - server=/ad.home.lan/172.16.1.24.172.16.1.23 (That doesn't work) neither does 172.16.1.24/172.16.1.23 Oct 06 22:48:51 h3sp4wn: as I wrote hours ago, use two entries with one ip in each Oct 06 22:50:11 xMff: Sound I must have missed it Oct 06 23:41:31 build #4 of adm5120 is complete: Failure [failed shell_3 shell_10] Build details are at http://tksite.gotdns.org:8010/builders/adm5120/builds/4 Oct 06 23:41:33 build #5 of avr32 is complete: Failure [failed svn shell_10 compile_12] Build details are at http://tksite.gotdns.org:8010/builders/avr32/builds/5 Oct 06 23:58:42 ping AndyI Oct 07 00:00:35 pong xMff Oct 07 00:00:35 xMff ping OK Oct 07 00:00:59 AndyI: #8044 is your ticket? Oct 07 00:01:14 xMff yes Oct 07 00:01:26 which cygwin version do you use? Oct 07 00:02:00 CYGWIN_NT-5.2 server00 1.5.25(0.156/4/2) 2008-06-12 19:34 i686 Cygwin Oct 07 00:02:08 so 1.5.something Oct 07 00:02:27 I am unsure howto decipher this version string :P Oct 07 00:02:49 I added a patch which implements a lame fix to just #ifdef the lockf calls out Oct 07 00:03:37 apparently its possible to emulate lockf() with windows _locking() api but I don't want to try that out yet, have not even a windows around Oct 07 00:04:17 alternatively you could try Cygwin 1.7.1 or newer Oct 07 00:04:41 let me know if the attached patch fixes compilation and whether you see further issues (esp. runtime) with opkg Oct 07 00:05:43 ok try apply patch Oct 07 00:06:47 RealOpty: I think I answered it myself, wanted to know whether old simpleshare tickets lingering around are still valid Oct 07 00:07:09 meh maybe one of them Oct 07 00:07:25 I didn't nuke the profile yet Oct 07 00:07:39 *profile ticket Oct 07 00:08:04 got a link on that one? Oct 07 00:08:40 can close this one https://dev.openwrt.org/ticket/7630 Oct 07 00:08:57 https://dev.openwrt.org/attachment/ticket/7647/SIMPLESHARE.mk Oct 07 00:09:19 and this one i dont care about https://dev.openwrt.org/ticket/7596 Oct 07 00:10:49 is it still valid? Oct 07 00:12:38 not sure. havent tested it. iirc after doing some research it was the wrong driver for the chipset or something Oct 07 00:13:10 this one is valid though https://dev.openwrt.org/ticket/7604 Oct 07 00:13:24 still seeing that issue Oct 07 00:13:42 ok Oct 07 00:15:50 https://dev.openwrt.org/ticket/7056 maybe u can close this. though its relevant but its fixed for others Oct 07 00:20:47 I will check this one tomorrow. Have to hit the bed now Oct 07 00:25:56 nite Oct 07 00:30:14 build #5 of brcm63xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/5 Oct 07 00:51:01 xMff: ping would you say this is a problem ? http://pastebin.com/G0LcNeX8 Oct 07 02:05:43 nbd: ping **** BEGIN LOGGING AT Thu Oct 07 02:53:27 2010 **** ENDING LOGGING AT Thu Oct 07 02:59:56 2010