**** BEGIN LOGGING AT Wed May 22 02:59:58 2013 May 22 05:53:33 build #244 of ixp4xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/244 May 22 08:28:15 jogo r36688 trunk/toolchain/gcc/Config.in * gcc: fix avr32 default GCC version May 22 11:38:57 jow r36689 trunk/package/network/config/firewall3/Makefile * firewall3: update to git head May 22 12:00:06 cyrus r36690 trunk/package/network/ipv6/6relayd/Makefile * 6relayd: Let OpenWrt override default CFLAGS May 22 14:17:06 jow r36691 trunk/package/network/config/firewall3/Makefile * firewall3: update to git head May 22 15:14:27 SpamBayes (in the issue tracker) is again convinced that I'm spam, but again doesn't give me the verification widget to prove otherwise. May 22 15:18:06 I'll reiterate my offer of US$500 for this bug: https://dev.openwrt.org/ticket/13337 May 22 15:18:24 and i'll add a comment to that effect on that bug when the system allows it May 22 15:24:48 brnt: do some dichotomy May 22 15:25:27 well, i started to, but decided to check my older builds first and discovered that i was incorrect. the problem was there, too. May 22 15:25:43 that's the comment i was trying to leave :-) May 22 15:44:53 brnt: someone needs to patch the RTL8366RB driver May 22 15:49:22 brnt: only for the kernel boot part ? May 22 15:49:30 or also the uboot boot part May 22 15:49:50 to do that right, need uboot to leave the switch either off, or fully separated May 22 15:50:00 then, the initial driver, leave it in the same state May 22 15:50:14 groz: yes May 22 15:50:16 and never have it actually passing traffic, until vlans are properly set up and tagged May 22 15:51:08 not familiar with that device, but, there should be a uboot config option to leave networking off on boot May 22 15:52:02 altho, if the switch powers up in a default 'pass everything' state May 22 15:52:12 then uboot will need to turn it off May 22 15:52:16 rather than just ignore it May 22 16:00:43 sorry, stepped away May 22 16:01:19 totally separated would be ideal May 22 16:01:34 the real problem is that dhcp clients connected to wired ports get upstream dhcp responses May 22 16:01:34 we disable the switch totally on our switches directly in the bootloader May 22 16:02:23 if the dhcp client never detects a link down, it won't get new dhcp from the router once the router finishes booting May 22 16:02:50 so if the ports are bridged during uboot and the link never goes down, it seems like the problem will still be there May 22 16:03:12 let me see if i can tell exactly when things get bridged... May 22 16:12:11 yeah, it looks like they're bridged during uboot May 22 16:12:52 look at the uboot env options May 22 16:13:00 there may be an option to disable network during boot May 22 16:16:06 here's the output of printenv: http://pastebin.com/cYU9m8jX May 22 16:16:21 the only thing that looks related is ethact=Eth0 May 22 16:19:23 merbanan: is there a simple flag in uboot for disabling the switch? May 22 16:21:25 what is the test_tftp all about, does that unit power up waiting on a tftp load initially ? May 22 16:22:05 brnt: no idea May 22 16:22:09 looks to me like it may be setup to do a tftp initially May 22 16:22:16 kinda like boot_wait on the old wrt54 May 22 16:22:59 not sure. it just gives the uboot menu, but the menu has tftp options, but it doesn't seem to do tftp by default May 22 16:23:08 (normal boot commences after waiting) May 22 16:25:31 merbanan: thx. do you know (more or less) what needs to be patched in the RTL8366RB driver? May 22 16:26:05 i'm not a driver guy and knowing what to ask for would be helpful May 22 16:27:07 well is the switch leaking from uboot start or only from driver load ? May 22 16:27:31 if it's leaking already during uboot, you need to change uboot so that it prevents leakage May 22 16:27:58 if it is driver load then disable all ports on driver init May 22 16:28:27 and enable them with swconfig later May 22 16:28:33 it looks like it's during uboot, so i'll start there. May 22 16:28:58 bootloader issues are more troublesome May 22 16:29:23 figures :-p May 22 16:29:45 but maybe using swconfig to disable all ports and then enable them will solve the problem you have ? May 22 16:29:53 possibly May 22 16:30:16 i tried ifconfig down/up, but the link never goes down and the clients don't re-request dhcp May 22 16:30:27 maybe swconfig will be more … linkish May 22 16:30:30 its a workaround but might solve your problem May 22 16:30:39 that's all i need :-) May 22 16:31:30 if it fixes your issue send the money to mr@loadsacach.net May 22 16:33:02 i'm not sure that workaround will close that bug :-) May 22 16:33:41 but mr loadsacach sounds reputable so i'll consider it :-D May 22 16:43:41 looks like swconfig taking down the link might be a no-go: https://forum.openwrt.org/viewtopic.php?id=40639 May 22 16:50:33 furthermore, `swconfig dev rt305x port 1 set disable 0` doesn't appear to do anything on my device May 22 16:52:31 no error, but no change to `swconfig dev rt305x port 1 show` or to LED and traffic still flows May 22 16:58:10 umm are you sure the name is rt305x ? May 22 17:05:57 that's how it's defined in /etc/config/network May 22 17:06:05 same results for `swconfig dev switch0 …` May 22 17:06:23 and any name other than those two gives an error May 23 01:02:48 cyrusff: did you intend to remove https://dev.openwrt.org/browser/trunk/package/network/config/firewall3/files/firewall.config#L102 with r36624? **** ENDING LOGGING AT Thu May 23 02:59:59 2013