**** BEGIN LOGGING AT Tue Jun 29 02:59:57 2010 Jun 29 03:09:15 build #51 of x86 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/x86/builds/51 Jun 29 04:17:43 build #53 of ubicom32 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/53 Jun 29 07:50:43 hello Jun 29 08:23:57 morning Jun 29 08:24:15 hello Jun 29 08:29:55 does luci touch the fstab locks, or is it only the block-mount code? Jun 29 08:30:15 fstab locks? Jun 29 08:30:35 the writing of /tmp/fstab is protected by locks around the writing code Jun 29 08:30:41 file locks Jun 29 08:30:49 in block-mount Jun 29 08:30:58 luci does not use file locking Jun 29 08:31:01 ok Jun 29 08:31:07 it also does not touch this file at all Jun 29 08:31:12 only /e/c/fstab Jun 29 08:31:27 right....so /etc/init.d/fstab is the culprit then Jun 29 08:32:08 someone is having a problem where there are double locks (i.e. lock not released before another lock attempted) Jun 29 08:32:58 are the locks using ash's (or is it busybox's) lock mechanism on the file or in memory? Jun 29 08:33:16 because I've been using them like semaphores Jun 29 08:33:29 you mean the lock command? Jun 29 08:33:32 yes Jun 29 08:34:20 it just exposed the flock() syscall which may act like a semaphore Jun 29 08:34:40 when acquiring the lock initially you can specify -w to wait for the lock to clear Jun 29 08:34:52 aahh...that's my problem Jun 29 08:34:56 I didn't have the -w Jun 29 08:35:10 but there wasn't a deadlock? Jun 29 08:35:25 there have been reports of occasional deadlocks Jun 29 08:35:31 hmm Jun 29 08:35:57 nested invocations of block_mount ? Jun 29 08:36:38 I think the hotplug + /etc/init.d/fstab might be the problem Jun 29 08:36:54 there's no nesting Jun 29 08:37:06 the code just locks, writes, unlocks Jun 29 08:37:19 but the lock doesn't have -w Jun 29 08:38:44 a while you're here... did you ever flash back to vendor firmware on the gw6000 ? Jun 29 08:38:57 I have yes Jun 29 08:39:14 I can't figure out how to do it with cfe Jun 29 08:39:37 you mean it doesn't flash, or it doesn't boot due to missing vfat filesystem? Jun 29 08:40:05 it doesn't flash, it rejects the image with error -4 for both the flashimage and the w command Jun 29 08:40:27 are you directly connected to the router (i.e. no gateway in between)? Jun 29 08:40:49 that cfe doesn't like gateways Jun 29 08:41:39 http://openwrt.pastebin.ca/1891655 should work, right? Jun 29 08:41:40 yep Jun 29 08:42:07 tftp also works, seems to be an image format problem with the vendor image [florian] gave me Jun 29 08:42:39 hmmm....do you need me to send you an image? Jun 29 08:43:18 actually I just need the srom embedded in the image Jun 29 08:43:24 its in /etc/wlan/ Jun 29 08:43:37 but unsquashfs fails for me Jun 29 08:43:48 neither v2, v3, v4 or -lzma work Jun 29 08:43:51 just a moment, let me see if I can get it Jun 29 08:45:27 I'm booting a GW6000 right now Jun 29 08:49:00 jow@openwrt ? Jun 29 08:49:50 pm Jun 29 08:51:40 got it and sent Jun 29 08:52:35 ok, regarding the lock Jun 29 08:52:39 lock -w only waits Jun 29 08:52:44 it does not acquire a lock Jun 29 08:53:07 so you'd do something like lock -w ... && { lock ...; do stuff; lock -u } Jun 29 08:53:09 ah, ok, so lock -w then lock ? Jun 29 08:57:52 so this http://openwrt.pastebin.ca/1891672 should work ? Jun 29 08:59:46 yes Jun 29 08:59:57 but you should cleanup $fstabnew within the locked section Jun 29 09:02:26 doh, right Jun 29 09:04:17 actuall this http://openwrt.pastebin.ca/1891676 should be right Jun 29 09:06:19 this looks okay Jun 29 10:07:47 cshore * r21987 /trunk/package/block-mount/files/block.sh: block-mount: Fixed mount deadlocks due to buggy use of locks in /lib/functions/block.sh Jun 29 12:52:08 has anyone experience in compiling own luci apps? i successfully compiled my app the first time... then i updated the html-template and created a new openwrt-image but my changes didn't apply :( i tried ./scripts/feeds clean, update and install, but the html-page is always the same :( Jun 29 13:33:18 qecko: if you want to modify the luci code manually you should checkout it separately Jun 29 13:34:31 qecko: e.g. svn co http://svn.luci.subsignal.org/svn/luci/branches/luci-0.9 ~/devel/luci and then src-link luci /home/devel/luci in /etc/feeds.conf Jun 29 13:35:06 jow_laptop: i did checkout separately... it worked after removing the complete luci dir: "build_dir/target-.../luci-0.9+svn" Jun 29 13:35:13 qecko: proceed with commenting out the default feed, then ./scripts/feeds uninstall -a -p luci; ./scripts feeds update; ./scripts/feeds install -a -p luci Jun 29 13:35:22 ah ok Jun 29 13:35:25 yes Jun 29 13:35:33 it is not rebuilt automatically Jun 29 13:35:49 what about "make package/feeds/luci/clean" ? Jun 29 13:35:53 make package/luci/{clean,compile} should help Jun 29 13:35:59 okay Jun 29 13:36:03 thank you Jun 29 17:13:45 build #61 of adm5120 is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/adm5120/builds/61 Jun 29 19:29:17 <[Fate]> in trunk, does crda compile? seems to require nl-tiny for me Jun 29 19:29:38 works for me Jun 29 19:34:52 <[Fate]> is it a package? Jun 29 19:35:06 <[Fate]> make clean didn't solve it Jun 29 19:35:45 libnl-tiny is a package, yes Jun 29 19:35:51 it's also used by iw, hostapd Jun 29 19:36:02 can you show me the error that you're seeing? Jun 29 19:36:53 <[Fate]> [...]/staging_dir/toolchain-mips_r2_gcc-4.3.3+cs_uClibc-0.9.30.1/usr/lib/gcc/mips-openwrt-linux-uclibc/4.3.3/../../../../mips-openwrt-linux-uclibc/bin/ld: cannot find -lnl-tiny Jun 29 19:37:17 with context Jun 29 19:37:21 in a pastebin Jun 29 19:37:21 please Jun 29 19:44:08 <[Fate]> let me distclean first, it didn't happen on a fresh checkout a few mins ago Jun 29 20:02:47 is it possible to use wildcards for conffiles (and if the wildcard includes a directory does it include sudirectories) in the Makefile? Jun 29 20:02:55 *package Jun 29 20:24:52 no Jun 29 20:41:06 <[Fate]> nbd: could you have a look at some of my tickets pls? Jun 29 20:46:14 nbd: well yeah they work like u said Jun 29 20:50:21 xMff: cool thanks we have a sysrq now Jun 29 20:50:24 <[Fate]> I get a compile error with libcc on a clean checkout Jun 29 20:50:34 <[Fate]> with local changes not touching libgcc Jun 29 20:50:34 OutBackDingo: does it help? Jun 29 20:50:37 <[Fate]> http://pastebin.ca/1892038 Jun 29 20:50:47 xMff: testing now Jun 29 21:06:23 <[Fate]> ok forget it, -static was in cflags -.- Jun 29 21:42:32 Anyone have any experience with the ar231x boards. I am having a few issues and was wondering if anyone could help Jun 29 21:43:00 <[Fate]> be more precise :) Jun 29 21:44:21 first teh board keeps rebooting every 5 minutes due to the watchdog timer and I need to know how to disable it. I also want to use the ath5k driver instead of madwifi, when I install the ath5k driver, the wireless interfaces are no longer available not is there any entry in teh networks config script Jun 29 21:45:19 I believe that teh ath5k driver is ported Jun 29 21:54:00 <[florian]> openwatts: it is not yet Jun 29 21:54:11 <[florian]> openwatts: ath5k does not work with ar231x at the moment Jun 30 00:10:41 nbd * r21988 /trunk/package/mac80211/ (20 files in 2 dirs): mac80211: update to wireless-testing 2010-06-28 Jun 30 00:10:46 nbd * r21989 /trunk/package/mac80211/patches/510-ar9100_tsf_preserve.patch: ath9k: fix TSF across hardware resets on AR913x Jun 30 00:29:39 build #52 of uml is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/uml/builds/52 Jun 30 02:01:34 build #57 of pxcab is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/57 **** ENDING LOGGING AT Wed Jun 30 02:59:57 2010