**** BEGIN LOGGING AT Sun Feb 06 02:59:57 2011 Feb 06 04:07:39 build #70 of ps3 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/70 Feb 06 05:07:25 build #82 of s3c24xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/82 Feb 06 09:37:35 moin Feb 06 11:07:22 did you hear? Debian's made Squeeze the stable release, finally! Feb 06 12:30:22 anyone on know if the brcm 96358 SoC has USB 2.0 or only USB 1.1? Feb 06 12:31:22 ehci Oops on me, but I don't know if that's because I was mistaken in thinking it had 2.0 or if it needs fixing Feb 06 12:31:46 or rather what it depends on from the platform code Feb 06 12:44:38 mb * r25380 /trunk/target/linux/omap24xx/patches-2.6.37/ (2 files): n810: Add more battery management code Feb 06 12:45:49 mb * r25381 /trunk/target/linux/omap24xx/base-files/lib/preinit/99_bme_pmm_image: n810: Add the CAL BME PMM extractor script to preinit. Feb 06 12:47:19 ping blogic Feb 06 12:54:08 cshore: it has 2.0 Feb 06 12:55:11 cshore: and it doesn't oops on me if I register the controller (but I can't really test, lacking a bcm6358 with usb connectors ;) Feb 06 12:56:43 KanjiMonster: thanks....any idea where to start looking to fix it? Feb 06 12:57:30 I guess it time for a debug compile and record the Oops Feb 06 12:57:39 yeah ;) Feb 06 12:57:57 now that I know it's a bug and not expected behaviour Feb 06 12:58:07 although you don't need to recompile, you can also find it out through the system.map in your build_dir Feb 06 12:58:32 Well I don't remember what build I have on the device Feb 06 12:59:01 so I'll do a fresh one Feb 06 13:04:06 tripolar * r25382 /packages/net/openssh/Makefile: openssh: update to 5.8p1 Feb 06 13:10:41 KanjiMonster: Ok, the 6348 is only 1.1 though, right? Feb 06 13:11:19 cshore: yes Feb 06 13:11:31 ok, and 47xx ? Feb 06 13:13:32 KanjiMonster: if you were to look for a manu that acutally supported open source who would you go with (router and/or nas wise?) Feb 06 13:13:57 *actually Feb 06 13:14:21 *manufacturer Feb 06 13:15:38 So far I'm thinking Ubiquity or Fon Feb 06 13:17:16 hm, good question Feb 06 13:17:54 Lantiq apparently does too, actually Feb 06 13:18:42 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/index.html Feb 06 13:20:24 cshore: I think you are now mixing device manufacturers and chip manufacturers; when talking about the latter, then I'd probably recommend latiq supported, ath79 or marvell kirkwood based devices Feb 06 13:22:15 ah, ok, yes I was confused....what about device manu? Feb 06 13:31:20 I'm thinking of replacing my sister's router (an asus -gu unit) with another router, so that I have a 47xx unit for testing (and if I'm going to buy a router, I'd rather put the money where I want it to go, which is to open source friendly manufacturer) Feb 06 13:40:45 cshore: well, asus belongs to those that are quite compliant in the way that they release gpl tars for every firmware version, not just the first one ever released Feb 06 13:40:57 its the same with netgear Feb 06 13:42:42 d-link, trendnet, tp-link need to be reminded to release sources to (newer) firmwares, and except d-link then replace the downloads, so you don't get the sources for older versions anymore Feb 06 13:43:40 what about device manufacturers that also use friendly chip manu (i.e. not the asus broadcom stuff) Feb 06 13:44:39 netgear is nice since they often have a populated serial header on the "bigger" devices Feb 06 13:45:50 netgear is funny though, they always pick the wrong platform for making an 'open source router' Feb 06 13:46:19 no matter which 'open router' you buy from them, you only get stuff that only works with binary drivers Feb 06 13:46:31 ah, ok, that's not what I want Feb 06 13:46:49 but some of their other devices which are not marketed as open source routers, work just fine with 100% free software Feb 06 13:47:01 oh, bizarre Feb 06 13:47:02 e.g. the wndr3700 Feb 06 13:48:13 the reason for their twisted view of open source routers is probably because they mostly look at stuff like tomato and dd-wrt Feb 06 13:48:34 okay, so that makes the list, netgear, ubiquiti, fon ? (ubiquiti being expensive, and fon doing the shared wifi thing) Feb 06 13:49:10 well, fon doesn't have any recent devices that work *properly* with openwrt yet Feb 06 13:49:18 oh, really? Feb 06 13:49:23 the ralink wifi driver still needs some work Feb 06 13:49:27 I though they were based on openwrt Feb 06 13:49:43 yeah, but fon is using proprietary drivers from ralink Feb 06 13:49:49 ah, I see Feb 06 13:49:58 recent netgear firmwares are also "based" on openwrt ;) Feb 06 13:50:18 <_trine> nbd, should madwifi work on the nanostation M2 Feb 06 13:50:22 so really I should look at wndr3700 or ubiquity Feb 06 13:50:23 _trine: no Feb 06 13:50:29 _trine: it probably never will Feb 06 13:50:43 <_trine> not at all even in b/g Feb 06 13:50:44 why? Feb 06 13:50:57 nbd: who? Feb 06 13:51:15 cshore: was talking to _trine. wndr3700 and ubiquiti stuff is fine Feb 06 13:51:17 i use both Feb 06 13:51:35 <_trine> I wanted to try it with madwifi because a have a script I have not converted which uses the madwifi commands Feb 06 13:51:55 nbd: any smaller netgear stuff you'd recommend (it's for my sister, and she doesn't need all the wndr3700 can offer) Feb 06 13:52:47 _trine: latest upstream madwifi supports the chip, but only in b/g mode, and it still wouldn't work on the board because it doesn't support not having a connected eeprom chip Feb 06 13:52:58 and even if it worked, it would be buggy as hell Feb 06 13:53:15 <_trine> ok Feb 06 13:53:39 cshore: the cheapest one i know (in europe) is the tp-link tl-wr741nd Feb 06 13:53:52 single stream atheros 11n wifi Feb 06 13:53:57 works well with openwrt Feb 06 13:54:25 can be flashed over the web interface, but doesn't have a recovery feature in the boot loader without serial Feb 06 13:54:35 nbd: I'm reading a couple of books on kernel devel, and am interested in trying to get the proprietary broadcom wifi working on brcm63xx (with multi-ssid). Any additional recommendations for reading Feb 06 13:54:55 the only reading recommendation that i have is the code ;) Feb 06 13:55:22 nbd: I'm looking for something with a relatively open source sympathetic device manu (e.g. netgear/ubiquity) Feb 06 13:56:01 well, i don't know any smaller netgear device that works well with openwrt Feb 06 13:56:10 and ubiquiti isn't really in the home access point market Feb 06 13:56:21 they're mostly doing outdoor equipment, enterprise APs Feb 06 13:56:35 heh, okay. When I get to the point of working on the code, can you help me out? Feb 06 13:56:55 nbd: ok, so what about asus? Feb 06 13:57:02 yes, but i don't have much time for looking into it myself, otherwise i would have fixed the bug already Feb 06 13:58:35 nbd; it'll probably more about questions having to do with never having seen broadcom datasheets or source code Feb 06 13:58:38 don't know any small well-supported devices there either Feb 06 13:58:55 i've never seen datasheets either Feb 06 13:59:00 and i try to avoid looking at the code Feb 06 13:59:06 it causes eye cancer Feb 06 13:59:12 nbd: :) Feb 06 13:59:21 heh, yeah, broadcom code is ugly ;) Feb 06 13:59:51 so about selecting hardware: i don't think going by perceived open source friendlyness is a good idea here Feb 06 14:00:01 nbd: I'm gathering that Feb 06 14:00:06 pretty much all of the home router manufacturers have both supported and unsupported devices Feb 06 14:00:13 as well as somewhat-open and not-so-open devices Feb 06 14:00:17 there really isn't much difference there Feb 06 14:00:25 so you might as well go with what's properly supported by openwrt Feb 06 14:00:30 right Feb 06 14:00:57 none of the manufacturers actually gives you the source code to *everything* running on the device Feb 06 14:01:06 so there is no ideal candidate here Feb 06 14:01:32 asus sometimes does with ralink devices ;) Feb 06 14:01:38 hmmm...ok. Feb 06 14:01:54 KanjiMonster: but i think even there they have some missing parts Feb 06 14:02:15 for some minor other components of the router firmware Feb 06 14:02:16 probably not intentionally, but there's sometimes the whole propriatary wifi stack included Feb 06 14:04:47 well that certainly clarifies things for me....and the suitation's no better at the high end? (i.e. when I buy for myself, at the RS Pro, or wndr3700 level)? Feb 06 14:04:58 bah, can't spell Feb 06 14:04:59 yes Feb 06 14:05:47 the wnr2200 looks like a good candidate for porting openwrt to it, it's basically the same hardware as the wnr2000v1, but has a proper bootloader that accepts real images like the wndr3700 Feb 06 14:06:08 what's inside? Feb 06 14:06:41 ar9132 with ar8216 (I guess) and ar9287 iirc Feb 06 14:06:54 32 mb ram/ 4 mb flash Feb 06 14:07:20 ar9132 with ar9287? that can't be Feb 06 14:07:31 ar9132 is a wisoc without pci Feb 06 14:07:48 okay, I should just guess chipnames ;) Feb 06 14:08:00 but iirc it was ar913x with ath9k ;) Feb 06 14:09:55 ah, it's an ar7240 with ar9287 Feb 06 14:10:25 could be Feb 06 14:10:32 at least that's what i found in a forum Feb 06 14:10:58 it is Feb 06 14:11:23 then its a wnr2000v3 with a proper firmware (and I just saw with usb) Feb 06 14:12:38 the wnr2000's should be avoided since they have stupid bootloaders Feb 06 14:35:34 mb * r25383 /trunk/package/base-files/files/etc/hotplug2-common.rules: hotplug2 rules: Run /etc/hotplug.d/firmware/ scripts (if any) on firmware load events Feb 06 14:43:09 mb * r25384 /trunk/target/linux/omap24xx/base-files/etc/hotplug.d/ (4 files in 2 dirs): n810: Add firmware generator hotplug scripts Feb 06 14:43:53 mb * r25385 /trunk/target/linux/omap24xx/base-files/lib/: n810: Remove firmware preinit scripts Feb 06 14:52:18 In which way/order, i.e. decending or accending, does the /etc/rc.d/KXXpackage get executed? Feb 06 14:53:31 ascending Feb 06 14:56:03 nbd: So, the /etc/rc.d/K99umount will be one of the last to get executed, right? Feb 06 14:57:36 yes Feb 06 15:00:21 ndb: I noticed that sometimes when I execute a 'reboot' on my Seagate DockStar through its serial-console or an ssh, the external USB partition wasn't cleanly umounted. When the LED on the device is off, I unplugged the external USB partition, connected it to my Linux desktop computer, ran a 'e2fsck /dev/sdb1', and e2fsck reported dirty inodes. Feb 06 15:01:09 maybe the umount isn't working properly Feb 06 15:02:04 ndb: I suspect that. Feb 06 15:18:57 nbd * r25386 /trunk/include/package.mk: allow packages to override/extend PKG_CONFIG_PATH Feb 06 15:41:52 obsy * r25387 /packages/net/transmission/ (Makefile files/transmission.config files/transmission.init): [packages] transmission: update to 2.20b4, add new options Feb 06 16:01:20 Makefile:45: *** Recursive variable `PKG_CONFIG_PATH' references itself (eventually). Stop. Feb 06 16:02:29 nbd: ^ Feb 06 16:03:10 odd, didn't show up here Feb 06 16:03:52 that's when I try 'make package//{clean,compile}' Feb 06 16:05:00 nbd * r25388 /trunk/include/package.mk: use := instead of = for PKG_CONFIG_PATH to prevent recursion Feb 06 16:05:01 that should fix it Feb 06 16:05:20 thanks Feb 06 16:06:13 np Feb 06 16:48:08 jow * r25389 /packages/libs/gettext-full/patches/005-uclibc-0.9.30-compat.patch: [packages] gettext-full: fix compilation against uClibc 0.9.30 (#8413) Feb 06 16:48:49 build #7 of lantiq is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/lantiq/builds/7 Feb 06 17:08:47 build #72 of pxcab is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/72 Feb 06 17:10:16 build #81 of brcm47xx is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/brcm47xx/builds/81 Feb 06 17:12:44 orbit crap Feb 06 17:14:41 jow * r25390 /branches/backfire/tools/ (Makefile cmake/): [backfire] backport r25264 Feb 06 17:17:54 jow * r25391 /branches/backfire/include/cmake.mk: [backfire] merge r25265, r25372 and r25376 Feb 06 17:17:54 can I trigger InstallDev without doing {clean,compile,install}? Feb 06 17:36:07 ,install is unnecessary Feb 06 17:36:49 other than that, you could remove staging_dir/target-*/stamp/._installed Feb 06 17:38:45 ok Feb 06 17:51:03 hm, it seems apps still link against openssl-0.9.8 even though it's 1.0.0 now Feb 06 17:51:24 make[3]: [clean-staging] Error 123 (ignored) hm.. Feb 06 17:51:49 rm -rf staging_dir/target-*; make Feb 06 17:52:07 so much collateral damage? :) Feb 06 17:52:23 should be enough to remove libssl, libcrypto.so Feb 06 17:52:26 and the ssl headers Feb 06 17:52:53 well, recreating the staging dir doesn't take much time Feb 06 17:53:00 and it'll get rid of everything that might be in the way ;) Feb 06 17:53:20 ah right, staging Feb 06 17:53:25 confused it with build Feb 06 19:00:45 nbd: how do the libnl and related patches look? Feb 06 19:01:11 haven't checked yet Feb 06 19:02:24 looks good Feb 06 19:04:09 I think I sent in a couple of other patches... some pretty trivial stuff. Feb 06 19:04:28 maybe I should get commit rights. :-) Feb 06 19:14:03 obsy * r25392 /packages/net/transmission/ (Makefile files/transmission.config files/transmission.init): [packages] transmission: add run_daemon_as user to default config, build for embedded system Feb 06 19:54:25 dingo * r25393 /packages/net/ (baresip/Makefile restund/Makefile): [patchteam] - fix building of baresip/restund - Signed-off-by: Alfred E. Heggestad Feb 06 19:57:05 nbd: weird, that doesn't fix it either. transmission is still linked against libssl.so.0.9.8 o.O Feb 06 20:00:51 did you also run make package/transmission/clean? Feb 06 20:01:03 after nuking the stating dir Feb 06 20:01:06 staging Feb 06 20:01:32 yes Feb 06 20:01:36 hm Feb 06 20:01:38 odd Feb 06 20:02:11 config.log doesn't pick up any specific version, just checks for >minimum_version Feb 06 20:02:20 what arch are you building? Feb 06 20:02:50 kirkwood Feb 06 20:03:11 with uClibc-0.9.30.1 Feb 06 20:03:36 and the 0.9.8 library did not reappear in the staging dir? Feb 06 20:03:43 correct Feb 06 20:03:55 hm Feb 06 20:03:59 no idea Feb 06 20:05:05 I don't get it either Feb 06 20:33:28 marek * r25394 /packages/net/ (batman-adv/ batman-advanced/): batman-adv: rename folder name to match project & package name Feb 06 21:07:12 mb * r25395 /trunk/target/linux/omap24xx/ (config-2.6.36 patches-2.6.36/): omap24xx: Remove 2.6.36 Feb 06 22:13:24 jow * r25396 /packages/libs/xmlrpc-c/patches/002-automake-compat.patch: [packages] xmlrpc-c: partly revert r25357 - it broke autoreconf (#8814) Feb 06 22:22:14 Is there a make variable that contains the STD include file locations Feb 06 22:22:24 ? Feb 06 22:22:53 you mean $(STAGING_DIR)/usr/include ? Feb 06 22:23:28 I guess, I will check to see if what I need is in there Feb 06 22:24:38 There is a usr/include in STAGING_DIR. Feb 06 22:25:09 There is no usr/include in STAGING_DIR Feb 06 22:25:49 I figure there must be a make variable that contains the include paths Feb 06 22:25:58 but I cant find it in the scripts Feb 06 22:51:28 nbd: even after a dirclean! how is that possible. it links agains 0.9.8 and 1.0.0: http://pastebin.com/8pSA66Sp Feb 06 22:54:14 loswillios: find build_dir staging_dir -name 'libssl.*' Feb 06 22:55:31 loswillios: and don't underestmate libtool :) Feb 06 22:55:43 there might be some *.la files lying around Feb 06 22:57:11 looks good to me: http://pastebin.com/9qrZr3bv Feb 06 22:59:16 maybe it picked up some crap from the host Feb 06 22:59:35 rgrep lbssl build_dir/target-arm_v5te_uClibc-0.9.30.1_eabi/transmission-* Feb 06 22:59:40 rgrep libssl build_dir/target-arm_v5te_uClibc-0.9.30.1_eabi/transmission-* Feb 06 23:02:26 hm, strings build_dir/target-arm_v5te_uClibc-0.9.30.1_eabi/transmission-2.20b4/ipkg-install/usr/bin/transmission-daemon |grep libssl says only 1.0.0 Feb 06 23:03:02 maybe some stale symlinks on the device? Feb 06 23:03:14 loswillios: then one of the dependant libs links ssl 0.9.8 probably Feb 06 23:03:28 ah, ldd pulls them in? Feb 06 23:03:36 libcurl is a likely candidate Feb 06 23:03:56 ah Feb 06 23:04:06 <_trine> I had this kind of thing earlier and did a make clean which has appeared to have succeeded in clearing it up Feb 06 23:07:19 that was it, thanks Feb 07 00:27:28 ah, nice. ubus is now properly handling sending and receiving events Feb 07 00:27:39 just need to add the pattern matching code for filtering events Feb 07 00:27:44 and then the code is pretty much feature-complete Feb 07 00:27:52 at least for the core functionality Feb 07 01:19:37 * xMff watches gitweb Feb 07 01:21:19 xMff: git question Feb 07 01:21:28 wrong person Feb 07 01:21:34 lol Feb 07 01:21:49 trying to figure out if i can switch the master tree i pulled from Feb 07 01:21:56 I usually need to lookup every single step in google plus use a tutorial Feb 07 01:22:33 OutBackDingo: you mean push/pull from a different remote, or make some remote the new origin? Feb 07 01:22:59 cshore: yes Feb 07 01:23:21 er, which one? Feb 07 01:23:42 make some remote new origin, as its a modified copies of the original Feb 07 01:23:51 you don't have to make the remote the origin to push/pull from it Feb 07 01:24:46 just push/pull from the a different remote (.e.g newremote/master instead of origin/master) Feb 07 01:25:31 after doing git remote that is Feb 07 01:26:22 cshore: hrmm looks like it is possible Feb 07 01:27:06 I think changing the origin can be done, but it's not as easy as just using a different remote Feb 07 01:27:23 You probably have to edit some file under .git Feb 07 01:34:17 build #77 of orion is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/77 Feb 07 01:36:09 jow * r25397 /packages/multimedia/ (3 files in 3 dirs): [packages] gst-plugins-base, gst-plugins-good, gst-plugins-ugly: use nls.mk Feb 07 01:45:39 cshore: on the other hand i could just blow away the build tree, and re-checkout from the patches tree instead of the master tree Feb 07 01:47:55 hrmm is there a specific directory structure for feeds ? Feb 07 01:48:06 no Feb 07 01:48:26 you can have arbritary number of subdirs Feb 07 02:51:40 yay, event pattern matching is done Feb 07 02:51:47 in ubus Feb 07 02:52:02 should be feature complete now Feb 07 02:52:34 that is needed for receivers to filter out unrelated stuff? Feb 07 02:52:53 yes, you can select what kind of events you want Feb 07 02:53:15 either with an exact match of the event id, or with a partial match (starting from the beginning of the id string) Feb 07 02:53:24 wildcard match is also supported Feb 07 02:53:42 ids are hierarchical? Feb 07 02:54:09 they should be made hierarchical, however the core does not enforce this Feb 07 02:54:12 you can pick any string Feb 07 02:54:40 ok Feb 07 02:54:53 some kind of dot notation? Feb 07 02:55:06 yeah, probably Feb 07 02:55:51 the tricky part for me was trying to figure out how to use the avl tree to optimize the pattern lookups Feb 07 02:56:55 not fully optimal yet, but at least it knows when it can stop going through the list because there are no more possible matches Feb 07 02:57:34 xMff: is uci dependent on lua ? Feb 07 02:57:39 OutBackDingo: no Feb 07 02:57:57 OutBackDingo: only at build time for the uci lua plugin Feb 07 02:58:21 so lua isnt needed for uci to do its job and wrt to complete its initialization Feb 07 02:58:29 right Feb 07 02:58:39 hrmmm Feb 07 02:58:44 * OutBackDingo wonders Feb 07 02:59:19 so essentially it could all be applied to a standard OS Feb 07 02:59:35 well uci by itself is not terribly useful Feb 07 02:59:43 xMff: i think the code of ubus also turned out well enough - most of it is high-level stuff **** ENDING LOGGING AT Mon Feb 07 02:59:56 2011