**** BEGIN LOGGING AT Sat Oct 27 02:59:56 2007 Oct 27 03:28:59 this is WIP day, right? Oct 27 07:12:16 juhosg * r9453 / (6 files in 4 dirs): [packages] 6tunnel: move init/config files to the right place Oct 27 08:05:34 hi, all! Oct 27 08:05:42 ejka, hi! Oct 27 11:59:58 I would like to welcome everyone to the October edition of WIP-Day ;-) Oct 27 12:00:40 * glp quickly runs back to the family-gathering :-) Oct 27 12:23:45 lo glp Oct 27 13:52:03 ejka, ping Oct 27 14:51:39 do anybody seen ejka today? Oct 27 15:22:06 glp: when is your gathering over (say, in UTC)? Hope all are having fun. Oct 27 17:39:39 dhr: UTC 12.00 sunday (24 hours after the beginning at UTC 12.00 today) Oct 27 17:40:25 glp: sorry, I meant your family gathering. Oct 27 17:41:12 dhr: a little later, the children are still running around in high-gear :-) Oct 27 17:51:54 nbd * r9454 /trunk/package/madwifi/patches/310-noise_get.patch: madwifi: don't poll the channel noise in a performance critical path Oct 27 18:09:00 http://pastebin.ca/751752 <- reneroger had the problem, same happens here when i use his config (http://pastebin.ca/raw/751616). i have no idea what's the propper value is for _FILE_OFFSET_BITS & ar7 Oct 27 18:12:11 crazy_imp: Busybox does not build with # CONFIG_LARGEFILE is not set Oct 27 18:12:20 crazy_imp: You need CONFIG_LARGEFILE set. Oct 27 18:13:20 Under "Advanced configuration options (for developers)" -> "Toolchain Options" set "Enable large file (files > 2 GB) support?" Oct 27 18:13:33 Dunno *why* busybox needs it, but it does. Oct 27 18:15:37 farnz: ok, it's selected by default, reneroger must have unset it. Oct 27 18:16:39 but if it's necessary for building busybox, we may should think about setting it static? Oct 27 18:32:17 crazy_imp: We should probably fix up busybox to build without it. Oct 27 18:45:03 farnz: guess it'll work to unselect it, if you don't want lfs for busybox fdisk and wc Oct 27 18:46:21 And on AR7, you rarely need LFS. Oct 27 18:55:52 farnz: not sure about that, some have usb, so you might connect a hdd Oct 27 18:57:24 Rarely, not never. Oct 27 18:58:56 farnz: why ? Oct 27 18:59:38 blogic: On most AR7s, you've got between 2MB and 16MB of flash, no USB HDD and no network file systems. On some, you'll need LFS. Oct 27 19:01:56 maybe someone should make two builds and compare the size of the packages to figure out how much space will maybe wasted by lfs :D Oct 27 19:41:22 blogic * r9455 /trunk/target/linux/generic-2.6/patches-2.6.23/065-rootfs_split.patch: fixed typo in generic squashfs split patch ... shame on you :) Oct 27 20:05:57 some devices (eg. Linksys WRV200) have GPLed source but are based on ucLinux and have some binary-only modules. Would it be reasonable to build something analogous to ndiswrapper to let ucLinux modules function on an OpenWRT kernel? Oct 27 20:09:12 dhr: what kind of modules? think it would make more sense to rewrite the module, rather than reusing the binary stuff Oct 27 20:15:40 http://wiki.openwrt.org/RTL8651BPort has a bunch of "stranded" devices. I haven't looked at what modules are binary-only. Probably the Airgo wireless. Maybe drivers for all the hardware features listed here: http://www.realtek.com.tw/products/productsView.aspx?Langid=1&PFid=11&Level=4&Conn=3&ProdID=70 Oct 27 20:27:29 22 GPIOs <- nice :) Oct 27 20:28:36 I have a WRV200. Great price for the hardware. Too bad about the firmware. No 3rd party firmware at this point. Oct 27 20:30:06 I was tricked into buying it (a year ago). The manual said it had a feature that I wanted. Turns out that it lied. (A feature of Openswan that they did not expose in the GUI.) Oct 27 20:37:12 I have a Motorola WR850G running Kamikaze 7.07. diag misidentifies my hardware. I've posted a fix in a trac entry (and mail) but nothing has happened so far. I don't really want to rebuild the whole system. Is there an easy way to patch a string in the firmware binary? Oct 27 21:56:32 dhr: i'm here at last ;-) Oct 27 21:56:57 exhausted? Oct 27 21:57:40 not to bad - it's around midnight and all the kids are happy and sleeping (+ i didn't get drunk) Oct 27 21:58:59 Sounds good! It's only 18:00 here (Toronto). Oct 27 21:59:31 how are things? Oct 27 22:01:05 I'm starting to look at my Motorola WR850 again (after last WIP day). Some mysteries. For example, it isn't doing NAT for me -- it's acting like a bridge. At least that's what I think right now. Oct 27 22:02:00 diag.c has a bug/deficiency that means that my router is mis-identified. I don't know if that can explain the behaviour. Oct 27 22:04:00 about the nat thing - did you amke any changes to the default config? Oct 27 22:04:43 I am using x-wrt 7.07 with no post-installation configuration to speak of. Oct 27 22:11:17 dhr: what kind of ip-adress do you get when you connect to the motorola? Oct 27 22:12:56 diag thinks I have a TEW411BRP+ router instead of a Motorola WR850G V2/V3 Oct 27 22:13:26 glp: when the WAN isn't connected, and I hook up to the LAN side of the router, all is well: Oct 27 22:14:23 I get assigned 192.168.1.183 (this time) and the router's LAN side is 192.168.1.1 Oct 27 22:16:16 what does the router state in the wan type of connection? when you'r connected Oct 27 22:16:18 (I'm doing testing as I report). Oct 27 22:17:13 how would you like me to ask? Use webif2 or some shell command? Oct 27 22:19:21 dhr: just try the webif Oct 27 22:19:56 x-wrt says, under network: WAN-LAN: Wan Configuration: Connection Type DHCP, interface eth0.1 Oct 27 22:20:24 I wonder why it isn't eth0? Oct 27 22:21:35 I have now connected the WAN interface and it does not seem to be picking it up. Last time I tried this, I power-cycled the router to get it to notice. That was crude. Oct 27 22:22:15 * dhr is going for dinner in a couple of minutes Oct 27 22:23:41 ok try to take a look at the network config file? Oct 27 22:23:55 ps says that there si a process udhcpc with args: -t 0 -i eth0.1 -b -p /rvar/run/eth0.1.pid -R Oct 27 22:24:39 what is the pathname of the config file Oct 27 22:25:07 /etc/config/network? Oct 27 22:25:30 oops: gotta run. BBL Oct 27 22:26:21 oki Oct 27 23:11:51 glp: I'm bacl Oct 27 23:11:56 glp: I'm back Oct 27 23:12:50 dhr: what does the network config file look like - try to past it to pastebin.ca Oct 27 23:20:47 http://pastebin.ca/752026 Oct 27 23:21:32 router -> notebook via scp; notebook to USB flash memory; flash memory -> desktop; desktop to pastebin. Oct 27 23:21:42 nbd * r9456 /trunk/package/madwifi/patches/310-noise_get.patch: oops.... committed an outdated patch Oct 27 23:21:55 I wonder why config interface lan says type bridge. Oct 27 23:23:45 dhr: the file looks completely standard Oct 27 23:25:19 dhr: the bridging aspect - the lan part of the router is several interfaces and they need to be 'bridged together' to talk to each other Oct 27 23:26:35 I powered up the router with no WAN and one LAN connection. It acts as expected EXCEPT when I later hook up the WAN, nothing appears to happen. Oct 27 23:27:06 ifconfig says that eth0.1 as transmitted 1543 packets but received none. Oct 27 23:27:53 the TX count is growing. I'll put a tcpdump on the other side... Oct 27 23:30:39 the part about why the interfaces are named: eth0.1 etc. -> because they are all part of the same switch Oct 27 23:31:31 dhr: but it really sounds like some thing is messed-up, somewhere Oct 27 23:32:19 dhr: and the router is offered a normal dynamic ip-adress on the wan side? Oct 27 23:32:58 the WAN I'm using is really a LAN (no surprise). I'm doing a tcpdump on the real LAN gateway and seeing nothing: tcpdump -i eth0 ether host 00:0c:e5:4c:e8:0c Oct 27 23:33:51 that mac address is what OpenWRT's ifconfig is reporting for eth0.1 Oct 27 23:34:30 dhr: yeah Oct 27 23:35:54 there is another scenario. If I power cycle the router with both WAN and LAN connections, something different happens. Let me try it again... Oct 27 23:36:13 dhr: and when you connect the wan port - and then connect to one of the lan ethernet ports with a cable, what then? Oct 27 23:39:33 continuing with my experiment first... the real gateway did see traffic on power cycle. Three bootp messages, My notebook then acquired an address on my real LAN (not NATed by the motorola router!). And it is connected to the internet. Oct 27 23:39:53 That sounds like the router is acting as a bridge. Oct 27 23:40:35 dhr: or as a dumb switch? Oct 27 23:40:51 I wonder if the WAN port is being treated as a LAN port and one of the LAN ports is actually being treated as the WAN port Oct 27 23:41:03 dhr: i'm thinking the same Oct 27 23:42:49 dhr: you can change the "layout" of the switch in the vlan configuration part of the network file Oct 27 23:45:01 dhr: or maybe try manually to test the other ports to see if one of them is wan? Oct 27 23:46:06 when I plug the WAN wire into LAN4 jack, things start to look good! Oct 27 23:47:03 so the numbering is wrong. Could that be associated with diag being wrong about what router I've got? Oct 27 23:49:00 dhr: it could i suppose (but i havn't thought much over this aspect before) Oct 27 23:50:01 dhr: but i've seen slight differences in the vlan config part between different router models Oct 27 23:50:27 dhr: like if you look at this: http://forum.openwrt.org/viewtopic.php?id=13164 Oct 27 23:51:58 dhr: this is a netgear router, and i wonder what would happen if you changed the 'option vlan1' into '4 5' instead of the present '0 5'? Oct 27 23:52:00 who sets up /etc/config/network? Oct 27 23:53:59 and: if I change this, am I likely to brick the router? Oct 27 23:54:35 try to read: http://downloads.openwrt.org/kamikaze/docs/openwrt.html#x1-80001.2.1 Oct 27 23:55:48 dhr: the part on 'switch' Oct 27 23:58:49 dhr: if boot_wait is turned on you can always reflash via tftp ;-) Oct 27 23:59:50 dhr: i assume the worst thing is that you might end-up having to find the wan port again Oct 28 00:01:36 boot_wait is on. The notes for this router says that it resets itself back to off after a reboot, but that seems to be due to the (now gone) motorola firmware. Oct 28 00:03:04 so: I'll change 1 2 3 4 5* to 0 1 2 3 5* and 0 5 to 4 5, right? Oct 28 00:04:29 dhr: try it and see what happens Oct 28 00:06:44 also: to get around the diag problem, I'm wondering if I can safely do a: nvram set CFEver MotoWRv203 (the current value is MotoWRv207). Is that likely to screw up the loader or would it have been only used by the rest of the firmware? Oct 28 00:07:54 alternatively I could nvram set MOTO_BOARD_TYPE WR_FEM1 (that variable is currently not set) Oct 28 00:10:37 dhr: i just have to take a look at my nvram reference source Oct 28 00:13:18 I posted a fix for this problem on September 29 but nobody has responded. I'd like to understand why. Oct 28 00:13:49 dhr: as i read the set CFEver command i think this just tells you what version of the CFE bootloader which is on the system Oct 28 00:16:27 diag.c line 716 tests CFEver to see if the platform is a Motorola WR850GV2V3. Do you think that it is safe to change the variable in nvram to trick diag into doing the right thing? Oct 28 00:17:20 [mbm]: ping Oct 28 00:19:17 dhr: i have no experience with doing such a thing - so i would classify it as experimental Oct 28 00:19:59 nbd: ping Oct 28 00:24:38 I set the variable, power cycled, and it came back with its old value! Oct 28 00:27:48 dhr: mmmhhhh Oct 28 00:28:57 I tried the other variable and its setting went away too. Is there a command I need to do to "commit" nvram changes? Oct 28 00:31:07 dhr: probably 'nvram commit' which saves nvram settings from memory to flash Oct 28 00:31:45 oh, that's useful to know. the nvram usage string does not list it. Oct 28 00:32:26 dhr: i have this some what sketchy reference manual in printed form Oct 28 00:33:01 dhr: this whole nvram business very strange for me Oct 28 00:35:23 Yahoo! nvram commit did the trick. Time to fix the nvram usage string. Oct 28 00:36:30 dhr: try it out Oct 28 00:37:30 I think that this only fixes lights. The port assignment is still screwy. Oct 28 00:39:13 but: the light for LAN4 corresponds correctly to the jack for LAN4. Even though the firmware thinks that LAN4 jack is actually the WAN jack. Oct 28 00:41:33 dhr: confusing ;-) Oct 28 00:44:36 I just editted /etc/config/network. Do I need to do anything to get it commited to flash? Oct 28 00:49:51 dhr: restart the network or reboot (it a standard config file system, so the files are stored + read when needed) Oct 28 00:59:48 WAN port now works as a WAN port! But webif2 seems to not be aware of this. Oct 28 01:02:32 dhr: super - and the time here just moved backwards, from 3.00 to 2.00 Oct 28 01:02:49 indeed :) Oct 28 01:03:13 googy US congress changed our "fall back". And I'm not even in the US. Oct 28 01:03:25 s/googy/goofy Oct 28 01:04:02 very odd to read a logfile which says: 02:59 and then next line begins with 02:02 Oct 28 01:04:32 it is now officially winter Oct 28 01:04:36 do you feel refreshed? Oct 28 01:05:01 dhr: not really - think i'll go and sleep for a while Oct 28 01:05:09 Winter starts ~ Dec 21 Oct 28 01:05:22 thanks for your help! Oct 28 01:05:58 I'm going to try to flash 7.09. I've never done that from withing OpenWRT. Oct 28 01:06:15 dhr: about the webif - it should actually change, in as much as it just reads what it says in the config files (maybe try to refresh the browser?) Oct 28 01:06:56 I tried that. Also: it didn't seem to ask me to log in again after a reboot. Odd. Oct 28 01:07:59 dhr: have you tried a force refresh where you wipe the browser cache= Oct 28 01:10:46 I guess that I didn't do it right. I exited firefox and did it again and it seems OK now. I used ctrl+reload; I guess it must be shift+reload. Oct 28 01:12:02 dhr: i always get those key-combinations wrong to Oct 28 01:12:34 * glp will go and sleep some hours **** BEGIN LOGGING AT Sun Oct 28 02:46:32 2007 **** BEGIN LOGGING AT Sun Oct 28 02:52:55 2007 **** ENDING LOGGING AT Sun Oct 28 02:59:57 2007