**** BEGIN LOGGING AT Thu Sep 27 02:59:57 2007 Sep 27 14:06:29 nbd * r9046 / (2 files in 2 dirs): do not fire up the qos-scripts hotplug part if the init script is disabled Sep 27 14:20:55 rwhitby * r9047 /trunk/package/kernel/modules/usb.mk: modules/usb.mk: Fixed the module autoload name for cdc_ether, and removed a typo from usbnet. Sep 27 15:02:21 juhosg * r9048 /trunk/target/linux/adm5120/base-files/ (6 files in 4 dirs): [adm5120] add initial diagnostic support Sep 27 15:02:43 florian * r9049 /trunk/target/linux/rb532/ (config-2.6.22 files/arch/mips/rb500/devices.c): Use the generic NAND driver Sep 27 16:36:04 How can I generate output from /etc/preinit to enable debugging of this? https://dev.openwrt.org/ticket/2400 Sep 27 23:37:23 When /etc/preinit hands over to /sbin/init (busybox binary) on boot, what happends between /sbin/init being called and /etc/init.d/rcS being called? Sep 28 01:23:19 nbd * r9050 /trunk/package/bridge-utils/Makefile: remove duplicate include statement Sep 28 01:23:57 nbd * r9051 /trunk/include/ (package-ipkg.mk package.mk prereq.mk quilt.mk target.mk): some minor fixes, cleanups, package build abstraction Sep 28 01:27:17 nbd; a while ago, you added a patch to speed up jffs (CONFIG_JFFS2_SUMMARY). I'm not sure if that was a response to this ticket https://dev.openwrt.org/ticket/2400 but I've added more info to it which suggests that it's not jffs related Sep 28 01:27:31 yes, i noticed Sep 28 01:28:31 I've been looking at the history of busybox related checkins and I can't seem to find anything that could be causing this Sep 28 01:28:46 IIRC; I was not seeing this in r7128 Sep 28 01:29:31 I'm also working on the assumption that /sbin/init will run rcS after reading the inittab Sep 28 01:32:24 I was looking at enabling "debugging aid" under the init section in the busybox config, but I have no idea what it does and its description is quite poor Sep 28 01:33:13 I also tried building a kernel without cpmac support, but I still see this same pause Sep 28 01:35:46 Can you think of any other tests I can run to help track this down? Sep 28 01:38:32 Also; is this platform specific to AR7? Sep 28 01:40:43 i've seen it before on a different mips platform Sep 28 01:40:45 amazon Sep 28 01:42:56 ok Sep 28 01:43:41 Do you think the evidence I've presented so far in the ticket, points to an issue with busybox's init ? Sep 28 01:45:13 nbd * r9052 /trunk/ (16 files in 16 dirs): add $(STAGING_DIR) as argument to the InstallDev template and update packages accordingly - this way we can reuse InstallDev to automatically generate UninstallDev or create -dev packages Sep 28 01:45:20 nabcore: not necessarily Sep 28 01:45:31 ok Sep 28 02:09:19 nbd; if /sbin/init is removed, there is no delay in the kernel "knowing about it"; https://dev.openwrt.org/ticket/2400#comment:5 Sep 28 02:09:59 more evidence (but not conclusive) that the delay is within the busybox binary Sep 28 02:10:26 makes sense Sep 28 02:11:29 I am not very familar with busybox's internals... what debugging options do I have open to me? Sep 28 02:12:01 no idea Sep 28 02:12:07 i normally use printf/printk debugging Sep 28 02:12:08 ;) Sep 28 02:12:11 :) Sep 28 02:13:33 lots oppertunity for those in openwrt/trunk/build_dir/mipsel/busybox-1.4.2/init/* ;) **** ENDING LOGGING AT Fri Sep 28 02:59:56 2007