**** BEGIN LOGGING AT Wed Mar 17 02:59:57 2010 Mar 17 08:08:21 ping kaloz Mar 17 12:26:26 AlexanderS, ping Mar 17 12:39:09 acinonyx * r20252 /packages/utils/tmux/Makefile: [packages] tmux: Update to v1.2 Mar 17 12:43:03 hi i'm using Openwrt Kamikaze, my app has 7 threads running, after some time 6 threads gets Zombie, can it be a memory issue Mar 17 13:16:06 augurel: Not likely. Are you sure it's something openwrt-specific? Mar 17 13:16:59 augurel: it is really hard to tell whats your issue without even knowing what the application does Mar 17 13:17:17 zombies can happen if the childs die and sigchld is not handled Mar 17 13:21:37 when i try to run my app on my local x86 machine, everything is OK Mar 17 13:23:01 mmm Mar 17 13:23:09 i've got a problem with opkg Mar 17 13:23:14 but on openwrt, when i started my app 7 pids appear( 1 pid for each thread) after a time 1 of them is alive(that is signal handler thread), and a zombie Mar 17 13:24:35 http://pastebin.com/pm1wraPq , this is my "ps -ef" output of my app Mar 17 13:24:57 augurel: Are you using the same threading library there? Mar 17 13:25:27 actually Not, on my machine i use pthreads Mar 17 13:25:46 as i know on openwrt, it is using uclibc Mar 17 13:25:51 Reordering packages before configuring them... Segmentation Fault Mar 17 13:26:45 Lallino: please tell me the exact opkg version in use and rn it under strace to the point where it crashes Mar 17 13:27:11 it uses early implementation of pthread, this is the reason of many pids for 1 app(1 pid for each thread) Mar 17 13:27:11 opkg version 0.1.5 Mar 17 13:27:16 but i'm on a special target Mar 17 13:27:20 so please let me check further Mar 17 13:27:44 ever recent opkg was 0.1.5 iirc, take the version from opkg list_installed | grep opkg Mar 17 13:30:49 opkg - 4564-3 - Mar 17 13:31:08 ok Mar 17 13:31:18 can i untar and copy the ipkg by hands? Mar 17 13:31:24 sure Mar 17 13:31:27 thx Mar 17 13:31:30 they're renamed .tgz archives Mar 17 13:31:38 do you want me to paste to you the full output? Mar 17 13:32:14 of the strace? You put it in pastebin Mar 17 13:32:20 *yes, Mar 17 13:35:27 i dunno have strace on the target yet Mar 17 13:35:30 sorry Mar 17 13:35:35 i will try later to get it Mar 17 13:36:01 i should put it on the target by hand since opkg not working ^_^ Mar 17 13:37:59 xMff: very strange, opkg does not work by ssh but it works by serial Mar 17 13:41:34 xMff : do u have any suggestion to how to identify the problem Mar 17 13:41:51 augurel: printfs, gdb Mar 17 13:42:31 if problem is my threads, why the management threads get zombie then? Mar 17 13:42:48 I think because the thread crashed and did not exist cleanly Mar 17 13:42:56 crashed due to a segfault most likely Mar 17 13:43:05 s/exist/exit/ Mar 17 13:43:47 segfault causes the crash of the program Mar 17 13:43:55 but 1 thread is still alive Mar 17 13:44:07 no Mar 17 13:44:27 a segfault must not crash the whole program Mar 17 13:44:36 if it is split into several childs Mar 17 13:45:20 ok, then checking for segmentation fault. Mar 17 13:52:56 rtz2: hi Mar 17 13:56:31 blogic: hi Mar 17 13:59:00 what does this mean? Multiple packages (freeswitch-minimal and freeswitch-minimal) providing same name marked HOLD or PREFER. Using latest. Mar 17 14:47:05 xMff : as usual, fault is a bug in my code Mar 17 14:47:14 ;) Mar 17 14:47:31 but thanks for the segfault warning Mar 17 14:47:43 I learned to not trust my C code even if it happens to work elsewhere Mar 17 15:07:38 lars * r20253 /trunk/target/linux/xburst/patches-2.6.32/ (4 files): [xburst] refresh patches Mar 17 15:11:35 does the BCM4718 support pci-express? Mar 17 15:21:28 nbd: looks like the sdk's Makefile needs to know about r20215's ARCH_SUFFIX addition, #6872 Mar 17 15:22:12 oh, right Mar 17 15:22:14 thanks Mar 17 15:32:20 * swalker wonders how much breakage a bulk INSTALL_DATA->INSTALL_CONF for config files would cause Mar 17 15:33:24 hmm Mar 17 15:33:30 install conf uses 0600 iirc Mar 17 15:33:45 could be problematic for programs with privelege dropping Mar 17 15:34:00 or maybe not Mar 17 15:34:14 but I'd wait until after the release with that Mar 17 16:03:10 nbd * r20254 /trunk/target/sdk/Makefile: sdk: fix up paths for $(ARCH_SUFFIX) Mar 17 16:58:49 nbd: did that fix https://dev.openwrt.org/ticket/6872 ? Mar 17 17:04:18 yes Mar 17 17:04:20 closed the ticket Mar 17 17:06:46 ty Mar 17 17:46:28 nbd: ping Mar 17 18:59:03 Hm, I did a custom brcm-2.4 build of kamikaze but network config /etc/config/network and /etc/config/wireless are not included. Any idea of what is missing in the config? Mar 17 18:59:52 Do I need to include "firewall"? Mar 17 19:01:40 both are generated on firstboot Mar 17 19:01:51 wireless by wifi detect and network by /e/i/netconfig Mar 17 19:04:52 hm, so for some reason that fails for me. Mar 17 19:11:42 Ok, I think it's a flash space issue. Mar 17 19:12:35 then its probably best to embed default ones Mar 17 19:16:12 Ok, yeah. That was it. I removed a few other packages to free some flash space and now it succeeds mounting the jffs2 and thus succeeds creating the configs Mar 17 19:16:46 I think default ones are not needed. You can't do much with the device anyway, if you can't config it due to missing jffs2 Mar 17 19:16:53 ah, thought you want to deploy an image w/o jffs at all Mar 17 19:17:04 no Mar 17 20:21:51 florian * r20255 /trunk/target/linux/generic-2.6/ (config-2.6.32 config-2.6.33 config-2.6.34): [kernel] add missing configuration symbols Mar 17 20:21:55 florian * r20256 /trunk/target/linux/malta/Makefile: [malta] include wpad-mini and mac80211_hwsim by default Mar 17 20:48:29 <[florian]> rtz2: ping Mar 17 20:49:39 [florian]: pong Mar 17 20:49:59 <[florian]> rtz2: http://openwrt.pastebin.com/Ckm0vDpn Mar 17 20:50:08 <[florian]> rtz2: with the rdc path you sent on the ml Mar 17 20:50:45 [florian]: that stupid error again Mar 17 20:50:51 [florian]: one moment Mar 17 20:50:57 <[florian]> np Mar 17 21:13:15 [florian]: could you give me the output starting from about 5 lines before that? Mar 17 21:15:21 <[florian]> http://openwrt.pastebin.com/Xj6DWv1p Mar 17 21:20:04 gahh, this was something incredible stuipd :/ Mar 17 21:20:37 [florian]: could you add echos for $kernel_size and $header_size before line 22? Mar 17 21:26:05 <[florian]> sure hold on Mar 17 21:29:44 <[florian]> kernel_size: 786176 Mar 17 21:29:44 <[florian]> header_size: 32 Mar 17 21:32:30 crap Mar 17 21:34:07 at least I found my bug Mar 17 21:43:43 [florian]: could you try with this calc_pad function? http://openwrt.pastebin.com/j0uVw5zW Mar 17 21:44:00 [florian]: there were a rather stupid mistake in it :/ Mar 17 21:46:08 <[florian]> rtz2: http://www.openwrt.pastebin.com/ZNFT7Ndj Mar 17 21:47:28 <[florian]> there is an image.tmp anyway: ---x--Sr-T 1 florian florian 768K Mar 17 22:45 image.tmp* Mar 17 21:54:46 <[florian]> rtz2: any clue? Mar 17 21:59:33 maybe Mar 17 21:59:50 one moemnt Mar 17 22:04:50 [florian]: found it Mar 17 22:05:07 <[florian]> rtz2: cool :) Mar 17 22:05:12 [florian]: line 113 of tools/firmware-utils/src/airlink.c Mar 17 22:05:39 the mode is 0x644 (hex), but I'm pretty sure it should be 0644 (octal) Mar 17 22:05:46 <[florian]> yes, I think you are right Mar 17 22:09:54 <[florian]> rtz2: I am still having the same error though Mar 17 22:10:32 <[florian]> rtz2: fixed Mar 17 22:10:44 <[florian]> rtz2: there was a spurious image.tmp file from last build Mar 17 22:10:47 ok Mar 17 22:11:34 <[florian]> rtz2: let me runtime test this now Mar 17 22:18:33 <[florian]> rtz2: the firmware is refused by the bootloader Mar 17 22:19:09 <[florian]> rtz2: let me dig out why Mar 17 22:20:20 <[florian]> rtz2: I will get off to bed now, but will see that tomorrow Mar 17 22:20:28 [florian]: ok, thanks Mar 17 22:20:53 [florian]: or one moment Mar 17 22:20:58 <[florian]> rtz2: yes? Mar 17 22:21:05 [florian]: do you have any kind of useable error message? Mar 17 22:21:09 from the bootloader? Mar 17 22:21:14 <[florian]> Error: invalid magic Mar 17 22:21:30 not very useful Mar 17 22:21:37 <[florian]> not really indeed Mar 17 22:21:54 <[florian]> but I think comparing two binaries and/or build logs should be helpful enough Mar 17 22:22:53 [florian]: if you send them to me, or upload them somewhere, I could take a look at it Mar 17 22:23:51 <[florian]> rtz2: ok, hold on Mar 17 22:30:28 <[florian]> rtz2: without your modifications: http://openwrt.pastebin.com/ztQ7xriH Mar 17 22:34:00 <[florian]> rtz2: this is with your changes: http://openwrt.pastebin.com/Mh7W0APw Mar 17 22:34:42 <[florian]> rtz2: this is an "old" image: http://alphacore.org/~florian/openwrt/rdc/openwrt-rdc-squashfs-ar525w.img Mar 17 22:34:59 <[florian]> rtz2: and a new one: http://alphacore.org/~florian/openwrt/rdc/openwrt-rdc-squashfs-ar525w.img-new Mar 17 22:35:12 <[florian]> hey, the problem is simple, the second image just lacks a kernel :) Mar 17 22:35:29 <[florian]> first one is 2MB, second one is only 1.3MB Mar 17 22:35:52 [florian]: bahhh Mar 17 22:36:15 found it Mar 17 22:36:27 in the makefile cat $(KDIR)/root.$(1) >> image.tmp Mar 17 22:36:36 cat $(KDIR)/root.$(1) >> $(KDIR)/image.tmp Mar 17 22:36:49 is ubnt.com down for anyone? Mar 17 22:37:03 I'm trying to find information on using the BDI3000 with routerstation/routerstation pro Mar 17 22:37:17 <[florian]> aport: it is down, indeed Mar 17 22:37:23 :( Mar 17 22:37:59 I'm using a cfg for the BDI3000 based on ap96... I can dump the flash but can't erase/program Mar 17 22:38:22 google cache has failed me too Mar 17 22:38:47 <[florian]> rtz2: this has to be done prior to calling airlink I think Mar 17 22:39:18 [florian]: no, that's by design Mar 17 22:39:54 [florian]: if I do this prior to calling airlink, the rootfs and the jffs2 start maker get included in the checksummed area Mar 17 22:40:05 <[florian]> and this will fail on next reboot Mar 17 22:40:13 [florian]: right Mar 17 22:40:29 [florian]: the checksum would have to get fixed on firstboot, which is rather ugly Mar 17 22:40:38 <[florian]> rtz2: yes Mar 17 22:41:00 hmm, looks like the BDI3000 doesn't support SPI flash Mar 17 22:41:06 [florian]: with some luck, the original firmware uses the filesize for flashing instead the length in the header Mar 17 22:41:32 <[florian]> rtz2: I think so Mar 17 22:41:53 <[florian]> rtz2: I am really off to bed now :p Mar 17 22:41:59 [florian]: good night Mar 17 22:42:11 <[florian]> rtz2: if you something between now and tommorrow morning I will be able to test before going to bed Mar 17 22:45:04 [florian]: i think it's the line cat $(KDIR)/root.$(1) >> $(KDIR)/image.tmp Mar 17 22:45:08 in the makefile Mar 17 22:45:22 [florian]: that's how it is now: >> image.tmp Mar 17 22:45:30 the $(KDIR) is missing Mar 17 22:46:21 [florian]: same thing on the next line with the mv Mar 17 22:47:35 [florian]: that's how it should be: http://openwrt.pastebin.com/XGcEiHZA Mar 17 23:07:51 does anyone know of a way to flash a routerstation pro via JTAG? Mar 17 23:43:06 lars * r20257 /packages/libs/ (4 files in 3 dirs): [packages] Fix linking of libsdl_{mixer,sound} Mar 18 00:54:34 lars * r20258 /packages/Xorg/lib/xcbproto/Makefile: Mar 18 00:54:34 [packages] xcbproto: Don't install python files Mar 18 00:54:34 The python files are not really needed right now and there have been built Mar 18 00:54:34 problems releated to them. So uncomment them for know. Mar 18 01:39:05 thepeople * r20259 /trunk/package/dnsmasq/files/dnsmasq.init: remove +1 as it isn't needed since 16619 Mar 18 02:20:38 nico * r20260 /packages/libs/rblibtorrent/Makefile: [packages] rblibtorrent: mark it as broken for now (see #6880) **** ENDING LOGGING AT Thu Mar 18 02:59:56 2010