**** BEGIN LOGGING AT Fri Mar 25 02:59:57 2011 Mar 25 03:33:36 If I wanted an initscript to start later in the startup process, how do I control that? Is there a way to make an initscript depend on another initscript finishing? Mar 25 03:47:27 antonlacon: you can try using system.foreground=1 and this will force all of the init scripts to run serially. Mar 25 03:47:35 that's how I get around certain issues. Mar 25 03:48:57 The issue I'm running into is that part of the system is on a usb disk, and the initscripts try using it before it's ready. Mar 25 03:49:16 ie a swapfile, for example. Mar 25 03:51:58 why are you swapping to a usb disk???? Mar 25 03:52:50 it's a swapfile, and I have room? Mar 25 03:53:02 system ram is higher priority Mar 25 03:53:16 and tmp is tmpfs, so if it needs the room, i'd rather it goto the disk than run the oom killer Mar 25 04:26:30 antoniacon: sounds like you should be using block-extroot to make your usb the root filesystem Mar 25 04:26:40 (well with overlay or whole disk, your choice) Mar 25 04:50:07 One step at a time. Taming minidlna at the moment. Thanks for the direction. Mar 25 06:19:33 antonlacon: it's easy and it will solve your race condition problem Mar 25 06:24:54 antonlacon: mkfs.ext4 /dev/sda1 ; mount /dev/sda1 /mnt ; cp -ar /overlay/* /mnt; umount /mnt Mar 25 06:24:54 after edit /etc/config/fstab so that Mar 25 06:24:55 config mount Mar 25 06:24:55 option target /overlay Mar 25 06:24:55 option device /dev/sda1 Mar 25 06:24:55 option enabled 1 Mar 25 06:24:55 option enable_fsck 1 Mar 25 06:24:56 ... etc (as per default /etc/config/fstab) Mar 25 06:24:56 if using trunk Mar 25 06:24:57 or Mar 25 06:24:57 option target /overlay-failed Mar 25 06:24:58 option is_rootfs Mar 25 06:24:58 instead of target /overlay Mar 25 06:24:59 if using backfire Mar 25 06:26:44 Changing course... Is there an easy way to control one of the kernel's build options through openwrt's menuconfig? I'm looking at inotify_user support. Mar 25 06:26:57 A sample that I could look at, or other doc would be great. **** ENDING LOGGING AT Fri Mar 25 07:03:19 2011 **** BEGIN LOGGING AT Fri Mar 25 07:04:14 2011 **** ENDING LOGGING AT Fri Mar 25 07:16:35 2011 **** BEGIN LOGGING AT Fri Mar 25 07:18:14 2011 Mar 25 14:01:35 When I did a 'ls -l staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/usr/include/' and 'ls -l staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/usr/lib/", they both are empty. However, both 'ls -l staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/include/' and 'ls -l staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/lib' a lot of files. Meanwhile, both CFLAGS and LDFLAGS point to both " Mar 25 14:01:37 staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/usr/include/' and "staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/usr/lib" directories. This doesn't make sense to me. :( Mar 25 14:04:44 nbd * r26287 /trunk/package/mac80211/patches/580-mac80211_fix_minstrel_ht_crash.patch: mac80211: fix a deadlock in minstrel_ht when a HT client without valid MCS rates connects Mar 25 14:04:49 nbd * r26288 /trunk/package/hostapd/files/hostapd.sh: hostapd: add more wps config methods to the config Mar 25 14:06:38 nbd * r26289 /branches/backfire/package/mac80211/patches/583-mac80211_fix_minstrel_ht_crash.patch: mac80211: fix a deadlock in minstrel_ht when a HT client without valid MCS rates connects (backport of r26287) Mar 25 14:21:46 build #88 of uml is complete: Failure [failed compile_10] Build details are at http://tksite.gotdns.org:8010/builders/uml/builds/88 Mar 25 14:32:31 When I tried to compile a program that uses some mathematical functions, i.e. atan, cos, sin, log10, sqrt, pow, etc., the compilation spitted out error messages saying "undefined reference to 'FUNCTION'" where 'FUNCTION' represents atan, cos, sin, log10, sqrt, pow, etc. Mar 25 14:35:07 Does that mean the libm does not support those mathematical functions? Mar 25 14:36:52 did you add -lm to the flags when linking the program? Mar 25 14:40:39 nbd: Yes. When I searched those mathematical functions (atan, cos, sin, and pow) in staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/include/math.h file and it doesn't have a reference to them. Mar 25 14:43:51 my libm has those functions Mar 25 14:45:35 nbd: under which platform? Mar 25 14:50:05 nbd: Arg, you must be looking at the 'staging_dir/toolchain-*/*-openwrt-linux-uclibc/include/c++/*/tr1/math.h' file while I was lookng at the 'staging_dir/toolchain-*/include/math.h' file. Mar 25 15:21:51 what's the best way to get a new image onto a running x86 box? Mar 25 15:24:48 doing a "dd" directly onto /dev/sda while it's running can corrupt the jiffs. Mar 25 15:30:51 philipp64|laptop: you could use kexec and then flash image from that instance Mar 25 15:37:26 build #88 of ar7 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/ar7/builds/88 Mar 25 15:53:09 nbd: I resolved the problem. Apparently, the Makefile from the package is missing the '-c' switch when it tried to just compile a source file into an object file. Hew ... Mar 25 16:39:08 build #111 of at91 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/111 Mar 25 16:40:30 nbd * r26290 /trunk/package/mac80211/patches/590-ath9k_fix_start_wakeup.patch: ath9k: fix chip wakeup on interface start - should fix some instability issues Mar 25 16:41:36 nbd * r26291 /branches/backfire/package/mac80211/patches/576-ath9k_fix_start_wakeup.patch: ath9k: fix chip wakeup on interface start - should fix some instability issues (backport of r26290) Mar 25 16:46:53 build #33 of lantiq is complete: Failure [failed shell_9] Build details are at http://tksite.gotdns.org:8010/builders/lantiq/builds/33 Mar 25 16:48:47 build #88 of avr32 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/avr32/builds/88 Mar 25 16:59:07 build #106 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/106 Mar 25 17:26:11 build #84 of ixp4xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ixp4xx/builds/84 Mar 25 18:06:33 nbd * r26292 /trunk/package/iptables/Makefile: iptables: libiptc.so is only a compatibility stub, split the package into libip4tc and libip6tc and adjust dependencies Mar 25 19:13:18 build #84 of kirkwood is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/kirkwood/builds/84 Mar 25 21:09:34 nbd: fyi, squashfs 4.2 linked against xz seems to build working images with compressed with xz as well as lzma (at least both booted fine on a bcm63xx device; the xz one was with 2.6.38, the lzma one with 2.6.37.4) - I now just have to figure out how to link only xz statically and not everything ;) Mar 25 21:14:00 ok Mar 25 21:14:21 and we need to figure out how to pass lzma compression parameters to xz Mar 25 21:14:36 I already have that figured out Mar 25 21:14:42 that was actually quite easy ;) Mar 25 21:15:19 ah, cool Mar 25 21:15:34 i'm currently working on de-bloating openwrt some more Mar 25 21:15:46 just figured out how to get rid of quite a few nasty cryptoapi dependencies Mar 25 21:16:09 nbd * r26293 /trunk/package/mac80211/ (21 files in 2 dirs): mac80211: update to wireless-testing 2011-03-24 Mar 25 21:16:32 nbd: http://pastebin.com/wzpbiEEu (<- broken tabs, so just as to show how I did it) Mar 25 21:17:35 i was talking about stuff like lc, lp, pb Mar 25 21:17:52 ah, ok Mar 25 21:18:30 since the default lzma parameters suck for most RISC architectures Mar 25 21:19:12 I see Mar 25 21:19:33 xz still has an api for that Mar 25 21:19:41 i just haven't looked much into how it can be used for that Mar 25 21:21:31 mksquashfs seems to expose some of it: http://pastebin.com/2EReLfYu Mar 25 21:21:53 ah Mar 25 21:22:05 does it have lc, lp, pb as well? Mar 25 21:22:12 or just dict size? Mar 25 21:22:14 nope, that's all Mar 25 21:23:55 nbd * r26294 /trunk/package/mac80211/Makefile: mac80211: fix zd1211 config override Mar 25 21:32:04 nbd * r26295 /trunk/package/mac80211/patches/300-pending_work.patch: mac80211: add a pending patch to remove the dependency on ecb Mar 25 21:32:24 nbd * r26296 /trunk/package/kernel/modules/crypto.mk: kernel: split up cryptoapi, move hash support and blockcipher/algorithm manager to separate packages Mar 25 21:38:17 build #81 of etrax is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/81 Mar 25 22:21:57 nbd * r26297 /trunk/target/linux/generic/ (2 files in 2 dirs): kernel: remove wrong chunks from the 2.6.37 and 2.6.38 version of the mips oprofile fix Mar 25 23:50:47 nbd * r26298 /trunk/target/linux/generic/ (4 files in 2 dirs): kernel: remove all local symbols from kernel modules at link time to reduce their size Mar 25 23:50:48 nbd * r26299 /trunk/package/Makefile: remove postinst files for preinstalled packages Mar 26 01:10:54 nbd * r26300 /trunk/ (25 files in 25 dirs): global cflags: remove -funit-at-a-time, it is ignored by recent compilers. instead add -fno-caller-saves, which improves compression ratio for generated code with no apparent negative effects on performance Mar 26 01:11:01 nbd * r26301 /trunk/target/linux/generic/ (2 files in 2 dirs): kernel: use -fno-caller-saves **** ENDING LOGGING AT Sat Mar 26 02:59:58 2011