**** BEGIN LOGGING AT Wed Oct 28 02:59:58 2009 Oct 28 10:52:04 sn9: did you end up finding the culprit? Oct 28 13:12:29 danage: seems some revisions either work or not, depending on something i have yet to determine Oct 28 13:14:35 sn9: are you trying to find your problem with brcm47xx Oct 28 13:14:41 yes Oct 28 13:20:55 sn9: so it's not the one patch that broke it? Oct 28 13:21:05 dunno yet Oct 28 13:21:23 man, florida with no uplink... Oct 28 13:21:38 wow and it's the 28th Oct 28 13:24:29 the uplink would be for the road, not for stationary use; roadrunner should suffice for that Oct 28 13:54:12 ok, i figured out why some revisions either work or not Oct 28 13:54:38 CONFIG_PACKAGE_kmod-ssb used to get randomly turned off, it seems Oct 28 13:56:34 jow * r18195 /trunk/package/lua/patches-host/100-no_readline.patch: [package] lua: add the 100-no_readline.patch to host build Oct 28 14:02:36 what kind of module is this Oct 28 14:08:19 {Nico}: you here? Oct 28 14:14:24 sn9: does this relate? https://dev.openwrt.org/ticket/4596 Oct 28 14:15:39 related, but possibly the opposite problem; checking Oct 28 14:17:36 ssb should be compiled directly into the kernel on the brcm47xx traget Oct 28 14:19:59 what is it Oct 28 14:31:02 ok, "svn blame package/kernel/modules/other.mk" squarely blames Hauke for this elusive goof Oct 28 14:31:23 r15197 Oct 28 14:32:17 the || should be && Oct 28 14:32:25 sherlock n9 Oct 28 14:33:24 reminds me of the fly in the typwriter in the film brazil. TINY error, huge consequence Oct 28 14:33:55 now, can somebody please commit the fix? Oct 28 14:34:53 ok Oct 28 14:34:56 sn9: what line ? Oct 28 14:35:29 @!TARGET_brcm47xx||!TARGET_brcm63xx Oct 28 14:35:30 ? Oct 28 14:35:35 yes Oct 28 14:35:40 should be && Oct 28 14:36:10 http://pastebin.ca/1646508 Oct 28 14:36:12 ? Oct 28 14:36:23 not sure what the disposition of the @ should be Oct 28 14:36:47 fixes breakage on brcm47xx, thx sn9 as a meesag e? Oct 28 14:36:54 does that describe the problem ? Oct 28 14:37:06 i'd ask nbd about the @, first Oct 28 14:37:10 nbd: ? Oct 28 14:37:25 he is on the train heading home, i just spoke to him on the phone Oct 28 14:37:30 oh Oct 28 14:37:31 it'll take a few minutes Oct 28 14:37:40 well, subway, not train Oct 28 14:38:05 who else can remember offhand such vagaries of the buildroot? Oct 28 14:38:16 florian Oct 28 14:38:36 it looks correct to me Oct 28 14:38:39 k; [florian]: ? Oct 28 14:38:43 lol Oct 28 14:38:44 :) Oct 28 14:38:47 {Nico}: Oct 28 14:38:50 hehe Oct 28 14:38:52 :) Oct 28 15:02:14 <[florian]> sn9: what ? Oct 28 15:03:00 [florian]: should that @ be there in blogic's patch, or not? or should there be two? Oct 28 15:03:57 only the one Oct 28 15:04:07 [florian]: http://pastebin.ca/1646508 Oct 28 15:04:41 <[florian]> what it does, is, we do not want the ssb module on brcm47xx or on brcm63xx Oct 28 15:05:43 [florian]: correct; Hauke intended that, but goofed Oct 28 15:06:49 <[florian]> roght Oct 28 15:07:45 @!TARGET_brcm47xx @!TARGET_brcm63xx Oct 28 15:07:51 that looks more correect ?! Oct 28 15:08:00 <[florian]> it does Oct 28 15:08:13 but should it be !TARGET_brcm47xx&&!TARGET_brcm63xx, @!TARGET_brcm47xx&&!TARGET_brcm63xx, or @!TARGET_brcm47xx&&@!TARGET_brcm63xx ? Oct 28 15:08:18 @!TARGET_brcm47xx @!TARGET_brcm63xx Oct 28 15:08:22 or what blogic just typed? Oct 28 15:08:25 <[florian]> or !(TARGET_brcm47xx||TARGET_brcm63xx) Oct 28 15:08:45 <[florian]> just like what it currently does with b43-common Oct 28 15:09:16 http://pastebin.ca/1646542 Oct 28 15:09:18 ok ? Oct 28 15:10:31 done Oct 28 15:10:31 blogic * r18196 /trunk/package/kernel/modules/other.mk: fixes DEPENDS of SSB module, thx sn9 Oct 28 15:11:49 thanks sn9 and blogic Oct 28 15:12:03 I have to leave now Oct 28 15:18:51 {Nico}: you here? we are hard testing the freeswitch packages :) Oct 28 15:18:58 stracing ! Oct 28 16:26:15 where is the code that handle the switch chip in the wrt350n-v1, again? Oct 28 16:26:30 sn9: what chip is it ? Oct 28 16:26:41 broadcom something Oct 28 16:26:48 in the kernel tree Oct 28 16:26:55 in tg3? Oct 28 16:27:02 hmmm Oct 28 16:27:09 i might be talking crap Oct 28 16:27:37 <[florian]> tg3 is the ethernet mac driver Oct 28 16:28:11 <[florian]> kmod-bcm57xx is probably what you are looking for Oct 28 16:30:36 [florian]: no, that's 2.4-only Oct 28 16:31:24 at some point, mb__ and nbd did some porting to 2.6 Oct 28 16:31:27 <[florian]> sn9: oh that's a pity Oct 28 16:32:04 wherever it is, it has an off-by-one error Oct 28 16:32:51 if i plug the cable into port 2, it tries to use port 1; if i use port 3, it tries port 2, etc. Oct 28 16:33:31 needless to say, ethernet only works at all when i nuke all vlans Oct 28 16:34:13 <[florian]> ah ah Oct 28 16:35:57 and only for a few mins; then it dies Oct 28 16:37:33 <[florian]> due to some watchdog or something different? Oct 28 16:37:37 yeah, there's a tiny bug which prevents it from working. But that should be easy to fix, if one is interested in getting it running Oct 28 16:38:02 It once worked well for me and appearently some tg3 upstream change broke it. Oct 28 16:38:10 Note that the switch does not work. Oct 28 16:38:47 The code for the switch on that device does not implement vlan, yet. Oct 28 16:38:58 That's maybe about 200 lines of code to add Oct 28 16:39:49 mb__: i thought nbd already did that Oct 28 16:40:08 so basically the driver for tg3 on ssb is in openwrt. It just has a few bugs. Oct 28 16:40:22 sn9: Oh, maybe. I don't know that. Oct 28 16:40:40 sn9: The off-by-one error is in that switch code then. Oct 28 16:41:01 do you remember the nature of the additional "tiny bug" you mention? Oct 28 16:42:01 mb__: ^ Oct 28 16:42:05 Well, it just doesn't work. There's two states a network device can have: 1) working 2) broken Oct 28 16:42:30 it works for a few mins when i nuke the vlans Oct 28 16:42:33 But as I said, it _did_ work for me some time ago and I did not change anything since then. So I guess an upstream change broke it. Oct 28 16:43:09 That may also be a switch problem. It's pretty hard to distinguish that Oct 28 21:45:51 _trine: you here? Oct 28 21:46:43 well i ask my question anyway maybe someone can help me Oct 28 21:46:53 i'm trying the freeswitch package for openwrt made by nico Oct 28 21:47:17 the freeswitch binary starts but then it says Aborted after a little while Oct 28 21:47:23 almost as soon as started Oct 28 21:48:09 i have done an strace Oct 28 21:48:12 here it is Oct 28 21:48:14 http://pastebin.freeswitch.org/10861 Oct 28 21:48:35 it's very strange because it seems it gets killed just after having created a pidfile Oct 28 21:49:13 i'm trying to tracking down the problem and i've seen that there could be a difference in the apache portable runtime library Oct 28 21:49:22 which is used by freeswitch to write files on disk Oct 28 21:49:38 here are the difference statically links at the library http://pastebin.freeswitch.org/10862 Oct 28 21:49:59 well i don't know where to go from here Oct 28 22:37:17 good night Oct 28 22:40:31 <_trine> yes Oct 28 22:50:32 <_trine> heres a link to the above pastebin Oct 28 22:50:35 <_trine> http://pastebin.com/m336b51ad Oct 29 00:10:18 to whomever can commit atm: the Makefile for usb-modeswitch puts doublequotes around $(TARGET_CC), which breaks ccache; please remove the quotes Oct 29 00:11:32 that's in packages/utils Oct 29 00:19:37 agb * r18197 /packages/utils/usb-modeswitch/Makefile: [packages] utils/usb-modeswitch: fix compilation w/ ccache. Thanks Daniel Gimpelevich **** ENDING LOGGING AT Thu Oct 29 02:59:57 2009