**** BEGIN LOGGING AT Mon Aug 08 02:59:58 2011 Aug 08 03:52:59 i've been itching to put wifi in these x86 boards... lookin for 2 cheap ones on eBay.. Driver wise I recall broadcom being funky.. what's got the good support with ability to do AP+STA? Aug 08 04:15:55 mute: atheros ar92xx or ar93xx Aug 08 04:16:25 dammit too late Aug 08 04:16:31 *JUST* got the broadcom one heh Aug 08 04:17:44 you won't be able to do ap+sta on that one Aug 08 04:18:06 oh well. was just trying to future-proof it. Aug 08 04:18:25 depending on which broadcom device it is, you might not even be able to do ap mode Aug 08 04:18:27 the boards have 2 mini-pci, 1 is already some atheros with 900mhz downconverter. Aug 08 04:18:51 really? hmm. Aug 08 04:18:57 it's a 43xx Aug 08 04:19:01 heh Aug 08 04:19:20 what's the 'xx'? Aug 08 04:19:22 bcm94318? Aug 08 04:19:34 they were $3.50 :D Aug 08 04:19:54 ok, that one might work with b43, but it's still crap Aug 08 04:20:41 get what ya pay for. so far my investment is $3.50 ... Aug 08 04:22:12 kids say it sucks at moms because no interwebz. it's 2 story so i don't know how to run ethernets around.. i just gonna have 2.4ghz AP downstairs and use the 900mhz to link the two, then 2.4ghz ap upstairs...? Aug 08 04:22:16 i dunno. Aug 08 04:22:50 oh, or bring them to Antarctica to get internet in the dorm ... hmm. Aug 08 04:24:17 all the ar92xx are pci-e :( Aug 08 04:24:33 there are pci cards with ar9220 as well Aug 08 04:26:12 well. i thought atheros was the best choice. then went back to bcm because i was certain i did repeater with mine before? Aug 08 04:26:24 nbd: so what are we going to do about the kmod-input-core issue? Aug 08 04:26:52 for now, I'm explicitly calling out kmod-input-core as s package, but I shouldn't have to. Aug 08 04:26:58 certainly will keep it in mind when i finally upgrade *my* network. Aug 08 04:27:36 got a wrt54g-series as my workhorse heh Aug 08 04:32:13 Alfa MB92M $30. AR9220 mini-pci "high power"... Aug 08 04:32:15 * mute goes to bed. Aug 08 04:32:18 thanks nbd Aug 08 05:23:08 or is it working as it should and I'm just not getting it? Aug 08 05:38:00 build #44 of mpc52xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/44 Aug 08 06:20:34 philipp64|laptop: I think nbd is on a plane to the US right now Aug 08 06:56:26 cshore: currently sitting at the airport Aug 08 06:56:45 :) Aug 08 06:56:51 ah the joys of flying Aug 08 06:56:52 nbd: what part of the states you off to? Aug 08 06:57:23 CO Aug 08 06:58:03 well, at least it shouldn't be /too/ cold just now. :P Aug 08 06:58:27 :) Aug 08 07:00:53 don't let them drive you mad :) Aug 08 07:01:52 right, off to work, bbl Aug 08 07:03:02 contract to work on wireless? Aug 08 07:03:19 ^@nbd Aug 08 07:04:08 openwrt related, not just wireless Aug 08 07:04:15 ah, ok Aug 08 07:04:55 proprietary project or completely open? Aug 08 07:05:55 the project itself is not open, but 90% of the work that i do on it is open and will be committed to openwrt Aug 08 07:06:28 and the rest is not worth opening up anyway Aug 08 07:07:34 that's cool....I'm hoping what I'm working on now will be the same way, but we'll see. Aug 08 07:08:04 i found it pretty easy to convince the company to do it this way Aug 08 07:09:31 just explain to them how much work they can save in the long run by getting all of the relevant core changes opened up and merged into openwrt Aug 08 07:09:49 I think for the parts that are based on openwrt that I've convinced the lead engineer, but we'll see what the project head says Aug 08 07:10:14 I haven't had a chance to talk about it with him yet Aug 08 07:26:40 nbd you still there? Aug 08 07:26:50 I have a sysupgrade question Aug 08 07:27:37 yep Aug 08 07:29:17 for checking the magic words and such it uses dd if="$1" .... The thing is afaict those bytes are used when actually writing the image, but I though dd would advance the stdin file pointer? I'm a little confused. Or am I mistaken and the magic words can safely be dropped? Aug 08 07:30:39 iirc it restarts reading the image Aug 08 07:30:44 so you should be able to drop the bytes Aug 08 07:31:04 nbd: ah, ok, I thought it was doing it with a single read...my bad Aug 08 07:32:10 I suppose because it's just a few bytes it's okay Aug 08 07:32:47 (I was thinking it was a single read for for http:// typle locations) Aug 08 07:50:46 does the kernel even let mtd refresh do what sysupgrade want to do? (reread the mtd partition table when a new image is flashed). AFAICT mtd partition table is only set on boot. And what exactly is mtd refresh for, really, if I'm right on that? Aug 08 07:52:39 aagh, look at the time...better get to bed if I want to have a hope of being awake tomorrow Aug 08 07:54:16 refresh only works with drivers that support it Aug 08 07:54:21 at the moment the only driver is mtdblock Aug 08 07:54:37 and it was only added, because the image write does not go through mtd Aug 08 07:54:55 so it's only used on x86 Aug 08 07:55:20 i'd like to get rid of this eventually, but right now i don't have a better idea on how to do sysupgrade on x86 Aug 08 09:09:39 build #66 of brcm47xx is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/66 Aug 08 09:13:56 build #65 of atheros is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/atheros/builds/65 Aug 08 09:18:14 build #64 of cobalt is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/64 Aug 08 09:22:10 build #63 of orion is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/63 Aug 08 09:37:13 build #62 of ps3 is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/62 Aug 08 09:42:36 build #66 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/66 Aug 08 11:29:50 nbd * r27930 /trunk/package/mac80211/patches/860-b43_restart_config.patch: b43: fix reloading config settings after channel changes or hostapd restart (#8033) Aug 08 11:32:15 nbd * r27931 /branches/backfire/package/mac80211/patches/860-b43_restart_config.patch: b43: fix reloading config settings after channel changes or hostapd restart (#8033), backport of r27930 Aug 08 11:58:48 build #68 of at91 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/68 Aug 08 13:23:46 nbd: i asked the manufactorer about that ar7240+ar9285 antenna switch problem we spoke about recently, and yes, they use the built-in antenna switching logic of the ar9285... Aug 08 13:24:32 implementing the antenna switching api doesn't seem like a big thing to me if one knows which register is used to control the switch internally (which i don't) Aug 08 13:24:35 any hints? Aug 08 13:25:01 kerneis * r27932 /packages/ipv6/ahcpd/ (4 files in 3 dirs): Aug 08 13:25:02 [package] ahcpd: upstream release 0.52 Aug 08 13:25:02 INCOMPATIBLE CHANGE: client mode configuration cannot be done in Aug 08 13:25:02 /etc/config/network anymore (at least for this release). It will Aug 08 13:25:02 be added back in the near future. Aug 08 13:25:03 kerneis * r27933 /packages/net/babel/ (4 files in 3 dirs): Aug 08 13:25:03 [package] babeld: upstream release 1.1.3 Aug 08 13:25:03 New option: -u (keep unfeasible routes), called "keep_unfeasible" in UCI. Aug 08 13:32:59 dangole: he's on a plane to the US atm i believe. Aug 08 13:34:03 EqUaTe: thx for the info Aug 08 15:59:16 nbd: speaking of x86 sysupgrade, I'm thinking of making sysupgrade images for x86 that a) for ext4 have a magic, size of kernel, kernel, and the rootfs image, and mounting the grub partition and replacing the kernel with the kernel in the sysupgrade image, and then replacing the ext4 rootfs with the new rootfs image (and then mounting it, and writing the saved configs) b) similarly for squashfs, but using mtd write -j for the squashfs instead of Aug 08 15:59:16 Of course since mtd refresh is supposed to work (it's doesn't, currently), I may not need to do all that for the squashfs images at least, and just fix mtd refresh. Aug 08 16:01:41 oh, nvm Aug 08 16:02:25 I don't need to do that for ext4 - I just need to get the kernel to reread the partition table on image write Aug 08 16:19:41 but using mtd write -j for the squashfs instead of <-- instead of what? Aug 08 16:29:15 EqUaTe: I stopped writing because I realized it was irrelevant Aug 08 16:29:25 heh Aug 08 16:30:07 unless your missing "just writing to the partition (because the goal is to be able to save configs, which is currently not possible)." because the client cut off before it shows on my screen Aug 08 16:30:26 yes. that's why i pasted that bit Aug 08 16:30:34 it got as far as 'squashfs instead of' Aug 08 16:30:49 ah, ok, too long lines Aug 08 16:30:54 yup Aug 08 16:31:58 bugger. now I'm not getting DHCP address but the network appears to be fine Aug 08 16:32:24 is dnsmasq running still? Aug 08 16:33:33 yup Aug 08 16:33:45 and I haven't changed the router anyway Aug 08 16:33:57 but my client is openwrt and I have changed it Aug 08 16:34:11 it's a client issue not server, I'm sure Aug 08 16:34:19 hairy muff. Aug 08 16:34:59 I've probably screwed something up Aug 08 16:34:59 maybe you've accidentally blocked the broadcast packets? Aug 08 16:35:15 flush iptables and change default to accept, see if it still works? Aug 08 16:35:18 I suspect it's ebtables that's screwing things up Aug 08 16:35:22 s/accept/allow Aug 08 16:35:25 ah Aug 08 16:35:35 yeah, that'd make sense. Aug 08 16:35:57 nope, it's all accept Aug 08 16:38:21 grrr...ok, let's try this with a 'clean' image and add back stuff to see what I've screwed up Aug 08 16:41:49 I bet it's related to trying replace lazy make evaluation of .config with immediate (replacing = with := in a copy of the .config and using that as the include instead of .config directly. Aug 08 16:42:20 trying to avoid make evaluating strings that I want verbatim Aug 08 16:43:24 heh Aug 08 16:44:16 make is evil; make is your only friend Aug 08 16:44:36 :P Aug 08 18:15:15 ? wtf....maybe it is the router... Aug 08 18:23:17 looks like dnsmasq is getting confused - it's not gone, but not doing the right thing after a while...will have to investigate tonight Aug 08 18:28:33 i've seen mine do that. inconsistently though, and generally when i haven't the time or motivation to investigate. restarting dnsmasq usually does enough. **** BEGIN LOGGING AT Mon Aug 08 18:30:38 2011 Aug 08 18:31:53 yeah, that's the problem with intermittent failures....it usually only fails when you're not able to look Aug 08 18:44:56 doh! it looks like I had an instance of dnsmasq running on the interface that was supposed to be receiving dhcp requests Aug 08 19:08:25 oops Aug 08 19:08:30 that'd definitely do it :) Aug 08 19:09:23 the weird thing with the issues i've had is that dnsmasq /appears/ to still be working. but iphones, ipads, and laptops all seem unable to renew their leases.. so it seems to be somewhat wifi related too.. but restarting dnsmasq fixes it Aug 08 19:21:49 ping nbd Aug 08 19:22:33 plane. Aug 08 19:26:31 EqUaTe: okay thanks Aug 08 19:26:51 he's on his way to the US Aug 08 19:27:01 it's possible he's arrived, but equally possible not Aug 08 19:27:13 iirc he was going to colorado Aug 08 19:42:17 EqUaTe: colorado? where in colorado? I'm in colorado... Aug 08 19:43:17 no idea Aug 08 21:10:14 any idea how to trigger a rebuild of the staging dir? or of just the host part of that? Aug 09 00:34:39 dirtyfreebooter: just arrived in broomfield **** ENDING LOGGING AT Tue Aug 09 02:59:56 2011