**** BEGIN LOGGING AT Fri Feb 04 02:59:57 2011 Feb 04 04:06:36 In the OpenWrt Build system is there a variable that contains the default include paths Feb 04 05:43:07 hi Feb 04 05:45:06 I have a question - how is it done that when running configure for the compilation of a package, the build system tells it to find includes and libs in a given place? Feb 04 08:11:44 build #69 of ps3 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/69 Feb 04 08:16:47 hcg * r25356 /trunk/target/linux/omap35xx/ (gumstix/profiles/gumstix.mk image/Makefile): [omap35xx] gumstix: create ubifs images Feb 04 08:53:33 ubifs... Feb 04 08:54:21 mtd look like ubifs Feb 04 09:22:49 <[florian]> KanjiMonster: pong Feb 04 09:34:28 gmorning Feb 04 10:02:40 build #82 of atheros is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/atheros/builds/82 Feb 04 11:26:14 * russell-- playing with ath5k in ap mode, not seeing a beacon from a potential client device Feb 04 11:27:20 oh, interesting Feb 04 11:27:40 iw wlan0 scan sees it, but network-manager doesn't Feb 04 11:28:01 or didn't Feb 04 11:28:03 nm Feb 04 11:28:41 nm can be laggy Feb 04 11:29:02 also it might still use wext instead of nl80211, not sure about that though Feb 04 11:35:26 mirko * r25357 /packages/libs/xmlrpc-c/ (3 files in 2 dirs): Feb 04 11:35:26 [packages/xmlrpc-c] Feb 04 11:35:26 - level up to svn rev 2080 Feb 04 11:35:26 - remove unneeded patches Feb 04 11:35:26 - add dummy package xmlrpc-c which is needed to avoid dependency Feb 04 11:35:26 problems when installing xmlrpc-c as package on the target later Feb 04 12:16:47 nbd * r25358 /trunk/package/mac80211/patches/560-mac80211_cmntr_drop_retransmissions.patch: mac80211: drop retransmitted data frames from the cooked monitor interface - reduces CPU usage Feb 04 12:16:55 nbd * r25359 /trunk/package/mac80211/patches/570-ath9k_bb_hang_check_fix.patch: ath9k: reduce false positives in the baseband hang check Feb 04 12:26:00 [florian]: reping Feb 04 12:51:05 jow * r25360 /packages/multimedia/gst-plugins-base/Makefile: [packages] gst-plugin-base: use nls.mk (for now) Feb 04 12:53:05 nbd: I noticed sometimes a few of this messages with latest mac80211 (ath9k, tp-link 1043nd): http://pastebin.com/WeZZYtJe Feb 04 12:53:48 jow * r25361 /packages/net/nbd/Makefile: [packages] nbd: use nls.mk (for now) Feb 04 12:58:50 jow * r25362 /packages/sound/scmpc/Makefile: [packages] scmpc: fix linking with openssl Feb 04 13:16:50 <[florian]> KanjiMonster: repong Feb 04 13:25:26 [florian]: the speedport w303v uses active low and active high gpios for its dsp, so I left it out for now Feb 04 13:26:40 <[florian]> KanjiMonster: allright, feel free to add the polarity fields if necessary Feb 04 13:26:47 originally I wanted to ask how I can find out where a wifi chip is attached, because I only saw the pci controller itself (or so I thought), but then I found out that bcm63xx doesn't have the controller as its own device (in contrast to ar71xx, to which I was used ;) Feb 04 13:27:44 the spw303v will be the first bcm63xx device with factory and sysupgrade images Feb 04 13:28:07 <[florian]> anything specific on this device? Feb 04 13:29:10 the cfe/stock firmware expects an image xor'd with "12345678" and un-xor's it before writing it to the flash Feb 04 13:29:18 <[florian]> funny Feb 04 13:29:41 also the crc32 for the image is calculated differently Feb 04 13:30:58 and the cfe allows braking in only when entering "$T&" at the "press any key to stop auto boot" prompt (needed to extract the cfe to find that out) Feb 04 13:32:59 so all in all this device is quite a PITA ;) Feb 04 14:26:16 build #81 of s3c24xx is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/81 Feb 04 14:40:15 KanjiMonster: quite impressing that you've worked that out though (re cfe) Feb 04 14:50:02 EqUaTe: I have a self built mips32 emulator which I modified enough to fake being a bcm5358 so that the cfe extracted itself into ram; then I just copied the ram contents into a file and threw objdump onto it and just tried some strings found it the image that looked like they might be the password ;) Feb 04 14:50:32 *bcm6358 Feb 04 14:53:21 nice :) Feb 04 15:03:41 EqUaTe: luckily cfe didn't seem to use any endianess dependend opcodes, since I only implemented those for little endian (since bcm47xx was my main target ;) Feb 04 15:05:42 KanjiMonster: always helpful ;) Feb 04 15:09:16 build #76 of orion is complete: Failure [failed compile_1] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/76 Feb 04 15:11:19 EqUaTe: if I someday decide to actually implement the uart(s), this emulator might even get useful ;) Feb 04 15:18:27 heh Feb 04 15:25:48 <[florian]> KanjiMonster: care to add bcm63xx support to qemu ;) ? Feb 04 15:29:22 hm. i got some bcm63xx hw around. whats supported on that soc? ethernet? uart? dsl? Feb 04 15:29:43 <[florian]> all but dsl and voip dsps Feb 04 15:29:50 <[florian]> I mean, with open source drivers Feb 04 15:30:00 <[florian]> someone could still use the binary drivers provided they have access to these Feb 04 15:30:14 hm. could dsl be 'made work' for recent kernels by using the binary stuff? Feb 04 15:30:36 <[florian]> someone having source for it could easily forward port to new kernels Feb 04 15:30:47 just thinking if the work to add more ram/flash would pay out Feb 04 15:31:08 8 or 16mb ram 2mbyte serial flash isnt enough for fun Feb 04 15:32:30 got multiple of these. were 4E a piece. adsl2+ modems. work better than the siemens stuff alice sent out Feb 04 15:48:42 roh that is spharion not siemens imho Feb 04 15:50:17 blogic: ah. well. siemens branded it ;) Feb 04 15:51:11 ok Feb 04 15:59:30 build #71 of pxcab is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/71 Feb 04 16:22:02 blogic: anyhow. they die after a year of usage. atleast ive seen that twice with that model Feb 04 16:24:41 sound like a cool unit Feb 04 16:24:42 :) Feb 04 16:25:34 [florian]: perhaps, someday. My emu is written in java, so plugging it into qemu isn't that easy - also the bcm63xx emulation is currently 99% "accept any writes and return the written values on read" except for one register or something (0xff410040) that the cfe checks for certain bits ;) Feb 04 18:43:07 KanjiMonster: is this all fun projects for you, or do you get paid to work on brcm63xx stuff? Feb 04 19:41:01 cshore: all fun Feb 04 20:06:20 KanjiMonster: best reason imo :) Feb 04 20:07:53 yeah, but to have the time for fun need to have a less demanding job .... Feb 04 20:10:29 you have the wrong job then :p Feb 04 20:12:46 cshore: or a less demanding wife. Feb 04 20:13:01 heh Feb 04 20:13:19 or a demanding job that is also fun Feb 04 20:13:20 ;) Feb 04 20:13:25 well, true Feb 04 20:13:27 yes, jobs that are fun are good Feb 04 20:13:40 hell, i really like my job.. and that's a big part of why i'm going to be working tomorrow. Feb 04 20:13:47 same here Feb 04 20:14:03 even if most of the stuff tomorrow is fairly basic. Feb 04 20:15:01 will be upgrading some core equipment at one of our sites, and rearranging the network there Feb 04 20:15:10 or rather, at one of the UKs main film studios Feb 04 20:15:53 i'm going to continue working on ath9k, ubus, and other openwrt related stuff Feb 04 20:16:00 :) Feb 04 20:16:20 if i'm lucky, i'll get a few hours in the afternoon to revisit the wnr2000 Feb 04 20:16:32 and i really must get around to sending that other one to KanjiMonster. Feb 04 20:17:03 nbd: how is the luci2 stuff coming along anyway? Feb 04 20:17:34 i haven't done any work on web interface related stuff yet Feb 04 20:17:37 but ubus is coming along nicely Feb 04 20:17:46 and i'll be working on a network interface daemon next Feb 04 20:17:53 what is ubus? not something i've come across before.. Feb 04 20:17:56 cshore: stalled till some other stuff is ready first Feb 04 20:18:07 nbd: does that include the init stuff? Feb 04 20:18:13 EqUaTe: a message bus, similar to d-bus, but much simpler and with a better API Feb 04 20:18:28 blogic: ok, you mean like ubus (core infrastructure stuff)? Feb 04 20:18:29 ah ok. i did wonder if it might be along that vein.. Feb 04 20:19:17 i want to make it as easy as possible to add an rpc api to any existing service with few lines of code Feb 04 20:20:04 any any service that gets such an api, will be accessible from the shell as well with no extra effort Feb 04 20:21:51 blogic: how's the init stuff comping along (what's it called uninit?) Feb 04 20:22:08 *uinit Feb 04 20:24:07 I'd like to rip out the preinit+init system and replace it with something better, but I know there's something already being worked on, so I've been waiting Feb 04 20:26:12 appsd Feb 04 20:26:17 it works Feb 04 20:26:23 waiting on ubus Feb 04 20:28:09 does it also replace hotplug to unify with the init system (since some init stuff depends on hotplugged devices and vice versa) Feb 04 20:28:11 ? Feb 04 20:28:24 at our university there's so much talk about 'innovation' and here it is happening (ubus, appsd, ...). this is great :) Feb 04 20:29:17 I think it will make init event driven Feb 04 20:29:24 its the only real solution Feb 04 20:29:29 right Feb 04 20:30:15 that's why I was asking about unification with hotplug (or hotplug triggers init events) Feb 04 20:32:11 Ok, here's a question: is it possible to build an initramfs that uses the squashfs for it's binaries, instead of having initramfs take up flash space that can't be used after the boot is done? Feb 04 20:33:16 xMff: dependency and event driven Feb 04 20:35:30 blogic: is there any way to set the core PATH (e.g. so that future events are triggered with external storage in the path, not only squashfs)? Feb 04 20:36:45 blogic: without making the external storage the rootfs that is Feb 04 20:37:44 ermrmm Feb 04 20:37:47 dunno Feb 04 20:39:53 cshore, iirc, the first segment in the path is parsed first, and if it hits, it'll ignore the rest Feb 04 20:40:55 maligor: right, but I'm wanting to extend to path in the case of no hit on the squashfs Feb 04 20:41:34 just append to the path environment Feb 04 20:41:58 aka: PATH=$PATH:/my/search/path Feb 04 20:42:26 maligor: yes, but the question is whether appsd allows you do that for a running appsd instance Feb 04 20:44:29 maligor: so that after running the even/command, appds will look on the new path, not just the path it started with Feb 04 20:44:59 maligor: and also will export the path to the processes it launches Feb 04 20:45:08 hack it to do it? Feb 04 20:45:33 maligor: my question is whether it already did it Feb 04 20:45:36 if you ask me, designs to do something better always last too long Feb 04 20:45:37 *was Feb 04 20:46:27 a bit like with git.. it's immensely practical but it doesn't look particularly beautiful Feb 04 20:46:46 maligor: I'll hack on appsd once it's usable Feb 04 20:47:11 yeah, it always depends on how fast, how much Feb 04 20:47:49 I suppose it's a blessing and a curse for working for a specialized software shop Feb 04 21:04:42 nbd * r25363 /branches/backfire/package/mac80211/patches/ (2 files): mac80211: merge fixes from r25358, r25359 Feb 04 21:04:44 nbd * r25364 /branches/backfire/target/linux/generic-2.6/ (2 files in 2 dirs): kernel: merge the eap frame drop fix for the bridging code from r25095 Feb 04 22:40:20 hmm.. why /etc/banner is kept on sysupgrade ? Feb 04 22:40:38 the revision is bad in the old file... Feb 04 22:42:25 Is it possible to make this keep-file-mechanism smarter? Like debian's? It silently replaces the file, if the user did not change it and displays a diff, if the user changed it. Feb 04 22:42:57 yea,or maybe keep default config in a gzipped file Feb 04 22:43:05 and make a tool to compare Feb 04 22:43:40 i mean the "new" default config Feb 04 22:44:19 mb__: would only work for squashfs Feb 04 22:44:26 but it wouldn't be able to display the new file Feb 04 22:44:35 only diff of old vs modified Feb 04 22:44:44 xMff: how about not treating /etc/banner like a conffile anymore? Feb 04 22:44:53 it seems to have a high false positive rate Feb 04 22:45:19 https://dev.openwrt.org/changeset/24742 ? Feb 04 22:45:32 oh, heh Feb 04 22:45:35 didn't notice that one Feb 04 22:45:36 ;) Feb 04 22:46:02 http://i.imgur.com/a19Nw.jpg Feb 04 22:46:04 yea... so why i keep it ? Feb 04 22:46:07 aka https://dev.openwrt.org/changeset/25066/ for backfire ;) Feb 04 22:46:29 Epsylon3: perhaps you upgraded *from* a version which didn't include the fix yet? Feb 04 22:46:50 no i dont think so Feb 04 22:47:01 it's still an opkg conffile Feb 04 22:47:06 i guess we should change that as well Feb 04 22:47:07 KAMIKAZE (bleeding edge, r25238) ------------------ Feb 04 22:47:08 blogic * r25365 /trunk/package/uboot-lantiq/files/include/configs/arcadyan-common.h: Feb 04 22:47:08 [uboot-lantiq] Feb 04 22:47:08 * tftp name was not in sync with that generated by openwrt Feb 04 22:47:27 nbd: ah, that makes sense Feb 04 22:47:28 its my actual banner... and just upgraded to r25359 Feb 04 22:49:12 blogic * r25366 /trunk/target/linux/lantiq/patches/ (101-header.patch 104-board_xway.patch): Feb 04 22:49:12 [lantiq] Feb 04 22:49:12 * fixes old ifxmips in header file Feb 04 22:49:12 * pass default ebu value to gpio driver Feb 04 22:50:02 i'm trying to access lan leds... on 88E6131 Feb 04 22:50:07 blogic * r25367 /trunk/target/linux/lantiq/ (3 files in 3 dirs): Feb 04 22:50:07 [lantiq] Feb 04 22:50:08 * adds ath5k platform support Feb 04 22:50:08 * adds several new profiles Feb 04 22:51:52 just got another WNR854T on ebay :) 20 euros Feb 04 22:51:54 :p Feb 04 22:52:12 i'm sure i can solder usb ports on that... Feb 04 22:52:40 http://wiki.openwrt.org/toh/netgear/wnr854t Feb 04 22:54:10 I have done an 'svn up' about an hour ago and now libs/xmlrpc-c breaks the compilation during configuration with this error messages "configure: error: conditional "am__fastdepCXX" was never defined." Feb 04 22:55:22 isofs: Unknown symbol load_nls (err 0) Feb 04 22:57:29 and maybe tomorrow a DIR855 Feb 04 22:57:37 with the OLED display :p Feb 04 23:00:29 i confirm... /lib/upgrade/keep.d/base-files-essential doesnt contains /etc/banner Feb 04 23:00:58 base-files neither Feb 04 23:05:28 jow * r25368 /trunk/package/base-files/Makefile: [package] base-files: remove remaining conffiles reference to /etc/banner Feb 04 23:06:08 nbd: it has no high false positive rate, just the fact that once it crept in it will be preserved forever until a full reset Feb 04 23:18:47 ^^ :) Feb 04 23:19:56 the original idea was to keep user modified banner files Feb 04 23:20:13 there is profile for that maybe... Feb 04 23:20:17 some use it as files for notices, in co-administrated setups Feb 04 23:20:30 -s Feb 04 23:21:56 i mean, they could change the /etc/banner source in profile Feb 04 23:21:58 xMff: ok Feb 04 23:22:16 like cat /etc/config/custom_banner Feb 04 23:22:33 too much Feb 04 23:22:33 xMff: dude Feb 04 23:22:40 dont paste that stuff i wrote with fon internals Feb 04 23:22:43 common Feb 04 23:23:04 :) Feb 04 23:43:44 is there some implementations of "long press" buttons ? like the original reset Feb 04 23:44:50 Who is mirko? Feb 04 23:46:40 a german guy from berlin :) Feb 04 23:47:10 Mirko did a r25357 this morning ([06:35] mirko * r25357 /packages/libs/xmlrpc-c/ (3 files in 2 dirs) and now libs/xmlrpc-c breaks the compilation. Feb 04 23:47:30 How do I downgrade? Feb 04 23:47:51 I meant to downgrade to r25356 Feb 04 23:48:32 If I do a 'svn up -r 25356', this will only downgrade the OpenWRT, but the libs/xmlrpc-c belongs to feeds. Feb 04 23:49:05 svn update -r 25356 -- /libs/xmlrpc-c Feb 04 23:49:07 cshore: you did implement the trx header fixup stuff? Feb 04 23:49:56 with a "." before /libs Feb 04 23:50:10 or go to the directory and svn update... Feb 04 23:50:44 Epsylon3: Thanks. I went to the directory and did a 'svn up -r 25356' to successfully downgrade. Feb 04 23:51:27 hmm its a feed... Feb 04 23:51:41 Yup. Feb 04 23:52:43 I tried to PM to mikro, but he hasn't responded yet. Perhaps, his location is in Germany and it's around midnight. Feb 04 23:52:56 1 AM Feb 04 23:52:58 here Feb 04 23:53:01 OK. Thanks Feb 04 23:53:31 and friday night :) Feb 04 23:53:38 good day to go out Feb 04 23:53:58 he hehe. Feb 04 23:59:54 xMff: yes Feb 05 00:01:21 cshore: is it normal that "TRX header not found" appears everywhere? Feb 05 00:01:37 xMff: no, that means something isn't working right Feb 05 00:01:38 I thought it was supposed to be only compiled for bcm63xx ? Feb 05 00:01:44 or 47xx Feb 05 00:01:55 people see it on ar71xx and it seems to work fine anyway Feb 05 00:02:28 hmmm.....I'll look into it Feb 05 00:03:22 ah Feb 05 00:03:28 obj.ar71xx = trx.o Feb 05 00:03:38 so its normal that it gets compiled in Feb 05 00:03:46 odd...I didn't realize that Feb 05 00:03:52 however it seems to be only for a few specifc boards Feb 05 00:04:21 https://dev.openwrt.org/changeset/24911 Feb 05 00:06:47 could it be an endianness issue for the TRX_MAGIC ? Feb 05 00:06:58 https://dev.openwrt.org/browser/trunk/package/mtd/src/trx.c?rev=24911 Feb 05 00:07:23 my point is that it is not needed on the wndr3700 at all Feb 05 00:07:33 in fact I wonder how it gets triggered at all Feb 05 00:07:51 if I interpret the mtd.c correctly it should only run if "fixtrx" was given on the commandline, right? Feb 05 00:08:01 you're right Feb 05 00:09:24 odd Feb 05 00:09:34 sysupgrade does not supply this arg anywhere Feb 05 00:09:56 fixtrx is only done once in a uci-defaults script for the wrt160nl Feb 05 00:09:58 it also happens for mtd with file Feb 05 00:10:11 hm? Feb 05 00:10:17 i.e. if you're including sysupgrade.tar.gz Feb 05 00:10:20 ahh Feb 05 00:11:12 so its purely cosmetic Feb 05 00:11:31 would probably be better ot print something like "no trx signature found, skipping fixup" Feb 05 00:11:36 yeah Feb 05 00:13:25 Since it's included by target not board, that's necessary because not all boards for ar71xx use trx (in fact probably only a few) Feb 05 00:13:35 only one, yes Feb 05 00:14:35 actually, here's a question: does that board need the fixup, or does it just happen to use trx? Feb 05 00:14:44 it needs it Feb 05 00:15:04 ok, so just changing the message makes sense Feb 05 00:15:05 there where quite some issues with the checksum on the wrt160nl in the past Feb 05 00:15:12 right Feb 05 00:18:39 time to walk the dog Feb 05 00:28:23 hmm that is weird, how can i have the good time on my router, rtc doesnt seems detected Feb 05 00:30:45 you mean why you have a correct time anyway? Feb 05 00:30:49 yea Feb 05 00:30:56 openwrt has some default rdate script Feb 05 00:31:03 which attempts to sync the time on boot Feb 05 00:31:15 i see... Feb 05 00:48:21 ok, lets see if the rtc is now working :) Feb 05 00:48:37 i make retro engineering from marvell sources (uboot) Feb 05 00:48:39 :) Feb 05 00:48:53 from netgear sources in fact Feb 05 00:57:35 ds1339 has 2 alarms nice :) Feb 05 00:57:43 dunno if that could help ;) Feb 05 01:32:19 mm no :( rtc-ds1307: probe of 0-0068 failed with error -5 **** ENDING LOGGING AT Sat Feb 05 02:59:57 2011