**** BEGIN LOGGING AT Thu May 31 02:59:58 2012 May 31 04:47:55 i just got x86 trunk. how come no iproute2 nor 'ip' in busybox? May 31 07:12:49 mute: iproute2 has been segmented into numerous packages for a more modular installation; opkg update && opkg install ip May 31 07:12:56 tc is its own separate package as well May 31 09:35:57 blogic * r32013 /packages/net/openvpn/files/openvpn.init: May 31 09:35:57 [packages] OpenVPN init script thinks client_disconnect is bool May 31 09:35:57 The OpenVPN init script incorrectly tries to use the client_disconnect option May 31 09:35:57 as a bool, but it's a string specifying a script to run when a client May 31 09:35:57 disconnects. May 31 09:35:57 fixes #11350 May 31 09:35:58 blogic * r32014 /packages/utils/lcd4linux/Makefile: May 31 09:35:58 [packages] fixes lcd4linux compile breakage May 31 09:35:59 mda166a does not exist May 31 09:45:25 blogic * r32015 /packages/libs/pthsem/Makefile: [packages] pthsem breaks on !glibc and !x86 May 31 09:48:50 blogic * r32016 /packages/libs/pthsem/Makefile: [packages] pthsem should depend on eglibc and not glibc May 31 09:53:21 build #23 of octeon is complete: Failure [failed shell] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/23 May 31 11:40:41 I have a virtual dedicated server thats idle May 31 11:40:56 would like to make a mirror or contribute some how :) May 31 13:15:12 heh, seems like dlinks beta 8.03 firmware for dir615 works OK with a PD on a dual stack pppoe May 31 13:15:46 but they choose to do a ppp0 ipv4 only and a separate ppp1 for ipv6 in their implementation May 31 14:46:11 i would like to implement bonding with netifd. could somebody give me some hints how to do that? May 31 14:48:04 build #22 of xburst is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/22 May 31 14:50:41 [Wed 2012-05-30 12:50:23 PM PDT] no swconfig support, remember May 31 14:50:41 [Wed 2012-05-30 12:50:27 PM PDT] <[florian]> oh right May 31 14:50:41 [Wed 2012-05-30 12:52:39 PM PDT] how do i talk to it without swconfig? May 31 14:51:09 mvswitch, that is May 31 14:52:37 <[Fate]> fresh backfire checkout (either 10.03.1 or head) does not build for me with ar71xx: cp: cannot stat `libgcc.a': No such file or directory May 31 14:52:50 <[florian]> DonkeyHotei: you just need to configure vlans, see how its done for atheros May 31 14:53:17 DonkeyHotei: you don't May 31 14:53:29 it applies a static configuration May 31 15:01:26 is the adm6996 series driver the same way, or can it use mvswitch? May 31 15:03:36 <[florian]> afair, it is the same May 31 15:11:57 nbd, [florian]: seems after a reboot, bootloader cannot use ethernet unless power is off first, even if eth0/1 are never brought up May 31 15:12:36 <[florian]> I suppose that the bootloader does not fully reset the mac and/or switch May 31 15:21:41 this is bizarre: when i try to send packets out eth0, in ifconfig, "RX bytes" increases but "TX bytes" does not May 31 15:22:00 <[florian]> probably because it assumes the link is down May 31 15:22:10 <[florian]> so it wont send any packets May 31 15:22:26 root@OpenWrt:/# [ 282.600000] PHY: cpmac-1:10 - Link is Up - 100/Full May 31 15:22:56 clearly not May 31 15:23:42 keep in mind, nothing it being sent to it from the other side May 31 15:23:47 *is being May 31 15:23:53 <[florian]> that does not explain why your tx counter does not increase May 31 15:23:55 <[florian]> it should May 31 15:24:01 <[florian]> unless the link is seen as down May 31 15:24:07 but why does rx increase? May 31 15:24:45 <[florian]> no idea May 31 15:24:46 rx increases only when i try to send May 31 15:24:54 and tx doesn't May 31 15:27:00 sending to the device also increases rx and not tx May 31 15:27:42 maybe cpmac is trying to process both sends and receives as receives somehow? May 31 15:28:25 <[florian]> or maybe there is a typo while updating counters? May 31 15:29:11 maybe May 31 15:29:19 what about eth1, do the counters increase there? May 31 15:29:44 no, and i already know eth1 is not used on this device May 31 15:29:59 ok May 31 15:40:16 my guess is that it's not a counter update issue May 31 15:40:33 probably the tx completion interrupt doesn't fire (which causes the tx counters to not get updated) May 31 15:40:36 and tx packets loop back into rx May 31 15:42:22 ah May 31 15:42:32 what would fire it? May 31 15:43:32 no idea May 31 15:44:34 are you saying that all it is is napi_complete() not getting called? May 31 15:49:10 nvm May 31 16:31:39 [florian], nbd: could this line be to blame? static int mii_irqs[PHY_MAX_ADDR] = { PHY_POLL, }; May 31 16:31:39 build #24 of ppc40x is complete: Failure [failed shell_15] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/24 May 31 16:33:20 <[florian]> DonkeyHotei: this just tells phylib to poll the PHY every 1s May 31 16:33:38 but it's an array May 31 16:33:53 <[florian]> and? May 31 16:34:07 only the first element is inited May 31 16:34:13 <[florian]> it's an array because you can have a phy at an address between 0 and 31 May 31 16:34:35 why is only element 0 inited? May 31 16:34:38 <[florian]> this is a C99 initialization May 31 16:34:48 <[florian]> means the complete array is initialized with PHY_POLL value May 31 16:34:55 for each element? May 31 16:35:30 <[florian]> yes May 31 16:35:43 <[florian]> it's equivalent to a memset May 31 16:35:44 i can't believe i didn't know that all these years May 31 16:36:00 <[florian]> it's a little tricky May 31 16:36:08 <[florian]> a loop initialization would be better and more conventional May 31 16:36:35 the cpmac irc does fire in general, but not for tx, i checked May 31 16:36:39 *irq May 31 16:37:09 what could be disabling it? May 31 16:37:18 <[florian]> napi May 31 16:38:37 where? May 31 16:40:19 <[florian]> actually the tx interrupt is never touched May 31 16:40:27 <[florian]> but the queues can be stopped in cpmac_end_xmit() May 31 16:42:58 looks like that function can only wake, to me May 31 16:44:00 <[florian]> yeah sorry, cpmac_check_status() can stop them May 31 16:44:36 <[florian]> I would suggest you raise the debug_level and see what's going on May 31 16:45:12 <[florian]> but certainly you should first properly configure your network with vlans 1 and 2 May 31 16:45:15 <[florian]> and see if that works May 31 16:45:25 <[florian]> and care about the counters later on with proper debugging May 31 16:45:55 both 1 and 2? May 31 16:46:29 <[florian]> just one should suffice, wan is the adsl anyway May 31 16:46:40 i did that; no effect May 31 16:46:48 hehe, guys congrats on trunk r32k :) May 31 16:46:59 <[florian]> do you have tcpdump running? May 31 16:47:06 <[florian]> what does ethtool say about counters? May 31 16:47:13 <[florian]> any undersized/oversized frames etc.. May 31 16:47:33 ethtool does not appear to be installed May 31 16:47:40 <[florian]> well install it :) May 31 16:50:28 making new image... May 31 16:51:04 <[florian]> bbl May 31 17:06:01 [florian], nbd: this changes ifconfig counters May 31 17:06:55 eth0: RX bytes:1095 (1.0 KiB) TX bytes:608 (608.0 B) May 31 17:07:07 eth0.1: RX bytes:0 (0.0 B) TX bytes:534 (534.0 B) May 31 17:07:30 what am i supposed to look at in ethtool? May 31 17:11:08 oh, bloody hell May 31 17:11:15 it was eth0.2 all along May 31 17:15:04 but it works only on ethernet port 1 May 31 17:15:42 I see that the mtd utility is used in some preinit/firsboot scripts... I have NAND flash, and mtd doesn't deal with bad blocks (which I have). Is there any chance for issues here? May 31 17:16:55 afaik, mtd is for NOR, so you might need to drop in a different script for your target May 31 17:20:09 well, the mtd utility is brought in as a default package. I can unselect it, but then I have issues with not switching to jffs2 during preinit. As long as the scripts aren't doing an erase or write, I think I'll be okay. May 31 17:20:29 or we could add bad block handing to mtd May 31 17:21:16 deselecting it != dropping in new scripts May 31 17:21:42 right, but I haven't had enough coffee to go hacking preinit scripts :( May 31 17:23:09 NAND flash isn't particularly uncommon, and it would be nice to have things like sysupgrade support. May 31 17:24:44 maybe nbd has a better idea May 31 17:46:06 build #22 of au1000 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/22 May 31 17:46:49 <[florian]> aport: nand can also be represented as a /dev/mtd block device May 31 17:47:36 Hi [florian], right now I'm using flash_eraseall and nandwrite on my /dev/mtd devices May 31 19:23:15 mirko * r32017 /packages/net/nodogsplash/patches/ (. 001-crash_on_missing_rdir_param.patch): [packages/nodogsplash] prevent nodogsplash from crashing when token/redir is not set - thanks to Moritz Warning May 31 21:08:19 [florian]: just in case you're still up, bringing up acx oopses the kernel, and i'm suspecting this might be because the mac address is aa:aa:aa:aa:aa:aa. ideas? May 31 21:09:20 <[florian]> no, I don't know where the mac address is supposed to be fetched from by acx May 31 21:10:33 from the card May 31 21:10:50 <[florian]> certainly, but I don't know whether it has a on-board eeprom for storing it May 31 21:10:51 but a lot of the vlynq cards have just dummy values there May 31 21:11:05 <[florian]> or whether it is supposed to be fetched from the main flash May 31 21:11:10 it likely has an eeprom that's unset May 31 21:12:07 the gt704wg has an env setting for "mac_ap" May 31 21:12:17 <[florian]> oh ok May 31 21:12:26 <[florian]> well then you are supposed to set the mac address manually using ifconfig hw ether May 31 21:12:28 it also has "macc" for the wan May 31 21:12:58 i tried that, but /sbin/wifi somehow reverts it May 31 21:15:05 <[florian]> I thik you can have an option macaddr in /etc/config/wireless which will set it accordingly May 31 21:15:15 <[florian]> start with a valid mac, then later we can figure out how to set it directly May 31 21:16:11 if i set that to anything other than aa:aa:aa:aa:aa:aa, /sbin/wifi thinks it's a different card May 31 21:17:51 also, usually the interface is invisible to ifconfig until /sbin/wifi brings it up May 31 21:18:14 so changing hw ether can't be done May 31 21:18:36 <[florian]> well you can still bring it up manually May 31 21:56:03 hrm.. and i get an error compiling sane-backends (technically the sane-libs package for hplip) May 31 21:56:14 the mkinstalldirs process fails May 31 21:57:04 http://ompldr.org/vZTFmag/pasta May 31 21:58:24 mystica555: looks like the -g is not recognized May 31 21:58:45 look at line 8 and then the call signature in line May 31 21:58:46 look at line 8 and then the call signature in line May 31 21:58:47 look at line 8 and then the call signature in line 9 May 31 21:58:57 * blogic needs a new keyboard May 31 22:02:45 indeed May 31 22:03:07 what gets me, is that it would appear that the script is provided by sane itself and isnt a systemwide thing.. May 31 22:03:21 so i don't understand why they would ship it in a broken state May 31 22:03:51 and, you call that a 'call signature'; i just consider that context sensitive help May 31 22:04:25 DonkeyHotei: set the macaddr in the wifi-iface section, leave the one in wifi-device as it was detected May 31 22:05:01 i don't see how a wrong mac address would oops the kernel though May 31 22:09:11 one way to find out May 31 22:14:37 nbd: ok, i did that, and it created the iface with the right mac, but then i ran /sbin/wifi and it reverted May 31 22:22:20 and then: May 31 22:22:23 [ 191.940000] Enabling firmware auto radio-recalibration. May 31 22:22:23 [ 191.970000] CPU 0 Unable to handle kernel paging request at virtual address 0000004c, epc == 942adc68, ra == 94aa48 May 31 22:23:00 hrm... so i tell portage to build sane, and it doesn't do anything visibly with that mkinstalldirs (or its masked somehow) May 31 22:27:05 hmm. a few month old alix2 build shows up no /dev/rtc **** ENDING LOGGING AT Fri Jun 01 02:59:57 2012