**** BEGIN LOGGING AT Mon May 16 02:59:57 2011 May 16 03:14:33 build #23 of ep93xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/23 May 16 03:40:11 Is K2.6.38.6 completely lost support of imq? May 16 03:41:00 I want to restore it in order to use Tomato's qos which can limit speed per ip. May 16 03:44:52 build #33 of orion is complete: Failure [failed compile_7] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/33 May 16 04:07:21 build #22 of mpc52xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/22 May 16 04:27:37 I made a theme for Luci and want to post. Should I send it to developers or post a ticket directly please? May 16 04:37:57 someone remind me who owns the mtd stuff? May 16 04:40:47 xeye0627, ticket May 16 04:40:52 Just post the patch to luci May 16 04:42:57 have to be diff files even they are all new files? May 16 04:47:10 Maybe May 16 05:15:20 RealOP, gzana. Thanks, a ticket is posted. May 16 07:32:44 build #33 of ps3 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/33 May 16 07:49:20 build #33 of pxcab is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/33 May 16 08:24:51 build #37 of at91 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/37 May 16 08:55:58 gzanan: afair you can limit with ifb as well May 16 08:56:04 gzanan: also with qos-scripts May 16 10:11:26 hey guys May 16 10:16:31 blogic: what's the status with ThinCan targets? May 16 10:17:23 blogic: I uploaded updated patch http://patchwork.openwrt.org/patch/989/ :) May 16 10:26:15 blogic: I also updated the patch for related tools http://patchwork.openwrt.org/patch/995/ May 16 12:23:47 Is automatic process group scheduling supported by default in K2.6.38 for openwrt? May 16 12:36:16 jow * r26910 /packages/net/aircrack-ng/ (3 files in 2 dirs): [packages] aircrack-ng: implement --ignore-negative-one option to cope with cfg80211 monitor mode channel issues May 16 13:31:47 lauri: ok May 16 13:31:48 looking May 16 13:32:17 ping jow_laptop May 16 13:32:19 hi May 16 13:33:43 jow_laptop: why did you patch aircrack-ng for the -1 channel bug? i fixed this bug by patchting mac80211 May 16 13:34:08 after i applied the patch everything worked as it should be - without patching aircrack-ng May 16 13:36:47 the patch is from here http://trac.aircrack-ng.org/ticket/742 May 16 13:46:55 the mac80211 patch has been rejected upstream May 16 13:47:02 so I will not touch the kernel May 16 13:47:18 jow_laptop: why was it rejected? May 16 13:47:30 and it works without patching the kernel with this option? May 16 13:47:34 yes May 16 13:47:59 okay nice May 16 13:48:05 thanks for the info May 16 13:48:09 np May 16 13:48:45 http://article.gmane.org/gmane.linux.kernel.wireless.general/61860 May 16 13:50:31 jow_laptop where did you find the patch the applied? May 16 13:50:57 the/you May 16 13:51:27 in #742 too May 16 13:51:48 that one was apparently added to aircrack svn as well May 16 13:52:47 ahh okay - seems i missed that one in 742 May 16 13:53:29 well it came later May 16 13:55:49 lauri: building a test image May 16 14:27:41 build #33 of lantiq is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/33 May 16 14:32:46 build #32 of x86 is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/32 May 16 15:58:26 I have updated to r26908. When I tried to compile freeswitch, the compilation calls to compile python first eventhough the compilation spits out a warning saying that python package isn't selected and would skip. May 16 15:58:55 Does that make sense? May 16 17:00:21 lauri: looking good :) May 16 17:30:21 good :) May 16 17:38:24 Why does python package not support threads? May 16 18:25:17 build #37 of s3c24xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/37 May 16 18:26:08 nbd: eh, you were faster than me ;) (saw your replies only after I sent mine) May 16 18:35:52 nbd: new fix... not sure why you wanted things moved to a later patch and not where the issue was introduced. May 16 18:37:57 well, i thought a different line was causing the const warning May 16 18:38:06 i only saw that 410 was clearly the wrong patch since it only moved one line May 16 18:38:31 the new one looks good to me May 16 18:45:23 nbd: btw, how does the ledtrig-ide-disk patch look? May 16 18:46:49 and if I wanted to get it upstreamed, any idea which kernel mailing list to use? I'm only familiar with stable and netdev. May 16 18:48:32 philipp64|laptop: try scripts/get_maintainer.pl from your kernel sources May 16 18:51:38 philipp64|laptop: haven't looked at it May 16 18:51:56 philipp64|laptop: haven't done much work last week, i'm still catching up on some important things May 16 18:52:12 so i don't want to do patch-monkey work at the moment May 16 18:52:30 is there a way to load a recovery initramfs with uboot ? with kernel and filesystem are broken... i tried to compile orion with CodeSourcery compiler ... i can load a kernel, but locks at init not found May 16 18:52:55 philipp64|laptop: there is a getmaintainer script in the kernel May 16 18:53:01 philipp64|laptop: it tells oyu the ML May 16 18:53:13 also that led patch should go upstream before being merged in owrt imho May 16 18:53:33 otherwise we have adriver in owrt, that people might base work on just to find that the final upstream fix is different May 16 18:54:23 i'm trying to setup a initramfs in orion tree... but i cant compile for the moment... May 16 18:54:40 /git/openwrt/openwrt/staging_dir/toolchain-arm_v5te_gcc-linaro_uClibc-0.9.32_eabi/lib/gcc/arm-openwrt-linux-uclibcgnueabi/4.5.2/../../../../arm-openwrt-linux-uclibcgnueabi/bin/ld: cannot open output file /git/openwrt/openwrt/build_dir/linux-orion_generic/base-files/libuClibc-0.9.32-rc2-git.so /git/openwrt/openwrt/build_dir/linux-orion_generic/base-files/libuClibc-0.9.32-rc3-git.so: No such file or directory May 16 18:55:18 blogic: ok... I'm just a little unsure about introducing the RCU dependency, but it looks like you can't build much of the networking code without RCU anyway, so it's pretty well entrenched at this point. May 16 18:55:42 Epsylon3: your tree is unclean May 16 18:56:02 yea, i will try again May 16 18:56:53 happy to have 2 things to do now... android cyanogen (motorola defy) and this openwrt May 16 18:57:54 already 2.6.38 hehe... my debian testing also May 16 18:58:28 LMDE (Linux Mint Debian Edition, x64) May 16 18:59:24 FOUI (Fucking old ubuntu installation, x86) May 16 18:59:45 i dont like recent ubuntu versions May 16 18:59:49 since 10.10 May 16 19:00:05 too much security enabled for a desktop edition May 16 19:01:04 its for android the x64 version... May 16 19:01:16 my ubuntu (for openwrt) is 32bit May 16 19:03:10 i found an interesting way to display unicode chars in busybox ls :) in 2 lines May 16 19:03:20 without any unicode lib May 16 19:05:39 https://github.com/tpruvot/android_external_busybox/commit/27a33d514c0fc12b06b99f31871330e67c27e9d6 May 16 19:05:41 you just have to disable the filtering, right? May 16 19:05:47 yea :) May 16 19:05:59 tpruvot = me May 16 19:05:59 this has been proposed before May 16 19:06:27 i sent a request to busybox but... May 16 19:06:37 for him ENABLE_UNICODE_PRESERVE_BROKEN is for input only May 16 19:06:44 ... May 16 19:06:49 not for output erf May 16 19:06:56 heh May 16 19:08:16 i need to ask for the "groups" applet... May 16 19:08:40 alias to "busybox id -G -n" May 16 19:08:51 WinSCP use that at logon May 16 19:39:07 erf 11 Go per repo :/// May 16 19:39:11 GB May 16 19:53:37 Epsylon3: I guess android? I noticed too that they weren't lying when they said android needs 20 gig for building May 16 20:07:03 <_trine> anyone fancy fixing arcrack-ng in openwrt so its a newer version May 16 20:08:06 <_trine> this is what you get when you try to use the svn of aircrack-mg in openwrt http://dpaste.com/543069/ May 16 20:11:01 xMff: last week, you and nbd acked the ieee80211d configuration patch, but it hasn't been committed yet. would you mind landing it when you get a chance? May 16 20:24:16 <_lore_> ping nbd May 16 20:27:20 _lore_: pong May 16 20:27:34 <_lore_> hi May 16 20:27:39 <_lore_> how are you? May 16 20:27:45 fine, how are you? May 16 20:28:10 <_lore_> fine tnx :) May 16 20:29:28 <_lore_> how are going a lot of stuff you are working on? May 16 20:29:47 making progress May 16 20:30:08 currently preparing the next mac80211 update May 16 20:30:19 <_lore_> cool May 16 20:30:42 <_lore_> I am still working on token channel access May 16 20:31:53 <_lore_> now the system is beginning to work May 16 20:32:04 cool May 16 20:32:18 <_lore_> but I am facing a boring issue May 16 20:33:09 <_lore_> the system often resets due to ath_tx_complete_poll_work May 16 20:33:27 <_lore_> and during the reset the throughput drops down May 16 20:34:31 <_lore_> I am using AR9280 chipset May 16 20:34:32 does it trigger because tx is really stuck? or does channel access take long enough to trigger the timeout? May 16 20:34:57 <_lore_> yes, I have commented the needreset condition May 16 20:35:07 <_lore_> and the system hangs May 16 20:36:25 <_lore_> the system is quite sensitive to the token tx/rx delay May 16 20:36:46 hard for me to comment on that without knowing the code May 16 20:37:43 <_lore_> you are right..I have no the right now...till tomorrow May 16 20:38:06 <_lore_> but I have observed the same issue also in csma/ca working mode May 16 20:38:23 hm, odd May 16 20:38:27 which mode? May 16 20:38:31 ap or sta? May 16 20:38:35 <_lore_> yes May 16 20:38:39 which one? ;) May 16 20:38:51 <_lore_> this issue is traffic depedent May 16 20:39:14 <_lore_> you have to inject 90Mbps of udp traffic May 16 20:39:33 <_lore_> one ap and 5 sta May 16 20:39:42 <_lore_> connected to the ap May 16 20:40:09 <_lore_> tomorrow i am going to carry out some more tests... May 16 20:41:54 <_lore_> have you ever faced this issue till now? May 16 20:42:47 haven't seen it yet May 16 20:42:55 <_lore_> ok.. May 16 20:43:36 <_lore_> do you have any ideas about which condition could cause the this hw stuck? May 16 20:44:44 no idea May 16 20:45:39 <_lore_> ok :) May 16 20:46:04 <_lore_> if I have news I will share with you... May 16 20:47:37 <_lore_> but just one more question May 16 20:48:23 <_lore_> why do we have to wait 1/2 second to reset the hw in ath_tx_complete_poll_work? May 16 20:48:47 <_lore_> and not in some other faster way? May 16 20:48:48 i think that's just an arbitrary interval May 16 20:48:58 making it too fast increases the likelihood of bogus resets May 16 20:49:02 e.g. if the medium is very busy May 16 20:49:55 <_lore_> ok May 16 20:51:27 <_lore_> and is there a maximum value for axq_depth beyond which the system is unstable? May 16 20:52:55 <_lore_> because if i limit the value of axq_depth to 3 or 4 the issue does not occur May 16 20:53:07 <_lore_> but the throughput is quite low May 16 20:56:11 does this happen with aggregation as well? May 16 20:56:16 or just with unaggregated traffic? May 16 20:56:31 <_lore_> yes...I am using aMPDU May 16 20:56:35 <_lore_> AMPDU May 16 20:56:43 ok May 16 20:57:56 <_lore_> if I stop the mac80211 pending queue when axq_depth goes beyoung 3/4 the system does not present the issue May 16 20:58:46 <_lore_> tomorrow I am going to try disabling AMPDU May 16 21:01:20 axq_depth shouldn't get big anyway May 16 21:01:25 since a-mpdu traffic is software-queued May 16 21:02:09 <_lore_> yes May 16 21:02:21 <_lore_> axq_depth is the hw queue depth May 16 21:02:40 <_lore_> so this is not due to AGGR May 16 21:03:43 <_lore_> I am wondering about the fact I inject burst of frame from mac80211in this working mode May 16 21:30:30 hi, i would like to install openwrt to my belkin router but http://wiki.openwrt.org/toh/start#belkin says openwrt supports a different version, is it safe to install to my router anyway? May 16 21:31:19 any dev around? May 16 21:41:29 build #37 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/37 May 16 21:44:18 build #38 of at91 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/38 May 16 21:44:33 Nihilus: i think we are talking about a totally different person;) May 16 21:45:49 ah May 16 21:46:23 I still like the guy May 16 21:46:55 I don't want him the be 'outed' or something May 16 21:49:57 nbd: got a question for you about a led driver I'm writing... http://fpaste.org/cGhY/ May 16 21:50:27 what is watch.S good for? hw-watchpoints? May 16 21:50:33 I'm seeing geos_led_init() being entered and exited with a good status (0), but the probe() function never gets hit. May 16 21:51:21 does the platform driver name match the platform device name in the arch code? May 16 21:51:32 markmentovai: oh, I will look soon, I promise, right now I'm too lazy to commit anything (again) May 16 21:53:09 nbd * r26911 /trunk/target/linux/generic/ (6 files in 6 dirs): kernel: backport some ssb changes to support a mac80211 update May 16 21:53:13 nbd * r26912 /trunk/package/mac80211/ (40 files in 2 dirs): mac80211: update to 2011-05-13 May 16 21:53:17 nbd * r26913 /trunk/package/mac80211/patches/570-ath9k_fix_adhoc_beacons.patch: ath9k: fix issues with ad-hoc beacon slot selection May 16 21:54:56 nbd: sorry, where does that get set? May 16 21:55:05 What is to prefer IO mmap Packets or not? May 16 21:55:27 philipp64|laptop: a platform driver only does something if it attaches to a platform device May 16 21:55:38 so somewhere, a platform device with the same name needs to be registered May 16 21:55:48 ok... so what am I doing wrong? May 16 21:56:34 do I need to get rid of the platform_device section? May 16 21:56:48 ? May 16 21:57:28 I was trying to do a mashup of the leds-net5501 and the leds-rb532.c modules. May 16 21:57:38 but clearly something is missing. May 16 21:57:52 yeah, what you're missing is an understanding of how platform device/driver stuff interacts May 16 21:58:25 they are normally used to allow arch code to register devices that driver code can attach to May 16 21:58:45 so that board/soc detection for that can stay in the arch code, separated from the generic drivers May 16 22:00:50 what's an example of the bits that I'm missing? May 16 22:01:31 if you used the rb532 stuff as reference, check out the rb532 arch code May 16 22:01:39 I was looking at the way that the leds-net5501.c driver (in .38) does things, but it seems a bit unclean. May 16 22:01:40 look for platform device registration for the leds May 16 22:02:02 gotta run to the bank. back in a bit. May 16 22:03:06 * nbd goes to get some sleep now May 16 22:04:20 final piece of advice: don't do cargo cult programming ;) May 16 22:04:33 just kidding May 16 22:04:44 don't even know what that is. May 16 22:05:21 http://en.wikipedia.org/wiki/Cargo_cult_programming May 16 22:06:22 xMff, got a few minutes for a chat May 16 22:06:49 RealOpty: just ask May 16 22:07:23 remember the simpletech simpleshare device that had a backup nvram May 16 22:07:31 yes May 16 22:08:52 would a proper fix to be add it to the partitions under static struct mtd_partition bcm47xx_parts[] = { May 16 22:09:05 yes May 16 22:09:18 but you have to ensure that the model detection is solit May 16 22:09:29 so that it does not influence other devices May 16 22:09:36 *solid May 16 22:10:00 indeed, thanks to the code that hauke added to support my netgear device that shouldnt be an issue May 16 22:16:10 xMff, i have a general idea of how to code the other parts i think the only thing im have trouble figuring out is bcm47xx_parts[4].offset May 16 22:17:45 bcm47xx_parts[4].offset = bcm47xx_parts[3].offset - ROUNDUP(NVRAM_SPACE, mtd->erasesize); May 16 22:17:54 xMff, does that look correct??? May 16 22:19:35 could make sense May 16 22:20:31 xMff, was the 'backup' nvram before or after the actual nvram??? i cant find my logs May 16 22:20:41 I don't recall May 16 22:21:18 run "nvram info" and tell me the offset May 16 22:21:39 I'd say the backup is the 2nd May 16 22:21:49 Offset: 0x00008000 May 16 22:21:57 hm May 16 22:23:13 well you want primary: May 16 22:23:13 bcm47xx_parts[3].offset = size - ROUNDUP(NVRAM_SPACE, mtd->erasesize) * 2; May 16 22:23:13 bcm47xx_parts[3].size = ROUNDUP(NVRAM_SPACE, mtd->erasesize); May 16 22:23:33 and secondary: May 16 22:23:33 bcm47xx_parts[4].offset = size - ROUNDUP(NVRAM_SPACE, mtd->erasesize); May 16 22:23:33 bcm47xx_parts[4].size = ROUNDUP(NVRAM_SPACE, mtd->erasesize); May 16 22:27:28 xMff, your saying that the code above is an example of what the fix should look like, yes? May 16 22:27:45 its how the partition should be laid out on the simpleshare May 16 22:27:56 3 would be the nvram and 4 the backup copy May 16 22:30:35 xMff, thanks, ill have a patch shortly. May 16 22:32:41 xMff, trying to fix this bug here: https://dev.openwrt.org/ticket/9323 May 16 22:33:00 i never thought that nvram would be relevant to openwrt May 16 22:33:07 but it turns out im wrong lol May 16 22:33:07 it is May 16 22:33:18 at least for mac addresses May 16 22:33:31 and for the wifi driver it might also hold vital callibration info May 16 22:34:18 in this cause its needed mainly cause of boot_wait May 16 22:42:20 xMff, what header is required for the function startswith ? May 16 22:43:22 its a private function in b44 May 16 22:43:31 introduced by a patch May 16 22:44:03 i dont think a header is required for inclusion actually May 16 22:44:06 buts very simple May 16 22:44:09 if (startswith(boardnum, "04FN")) <- example May 16 22:44:20 static inline int startswith (const char *source, const char *cmp) May 16 22:44:20 { May 16 22:44:20 return !strncmp(source,cmp,strlen(cmp)); May 16 22:44:20 } May 16 22:45:38 found it in the original source of my example too May 16 22:45:40 static inline int startswith (char *source, char *cmp) { return !strncmp(source,cmp,strlen(cmp)); } May 16 22:45:50 ty May 16 22:58:26 xMff, heres what i got for the detection so far : http://paste.ubuntu.com/608708/ May 16 22:58:29 what do you think?> May 16 23:00:20 it looks okay May 16 23:06:28 xMff, if (get_router() = ROUTER_SIMPLETECH_SIMPLESHARE) <-is that possible and correct format? May 16 23:06:41 yeah May 16 23:12:32 xMff, does my patch look correct and sane ???? http://paste.ubuntu.com/608724/ May 16 23:16:29 not quite May 16 23:16:39 you can't just create part 4 out of thin air May 16 23:16:51 you have to make place in the array for it May 16 23:17:12 xMff, inst that what i did in line 25 ??? May 16 23:17:48 yes but I am not sure whether it will appear on any other unit as well then May 16 23:18:06 and how well the stuff copes with a partition having no size May 16 23:19:47 indeed. May 16 23:20:05 I'd say just add a 2nd { name: NULL } entry May 16 23:20:47 then below, when initializing part 4, add a bcm47xx_parts[4].name = "nvrambackup" May 16 23:21:31 ahh May 16 23:21:44 oh and it might make sense to do bcm47xx_parts[4].mask_flags = MTD_WRITABLE; May 16 23:21:58 to make the nvram backup r/o May 16 23:22:40 readonly? May 16 23:24:10 xMff, like this { name: NULL, offset: 0, size: 0, }, ? or like you said??? May 16 23:25:04 line 26 May 16 23:25:21 kk May 16 23:25:42 { name: "nvram", offset: 0, size: 0, }, May 16 23:25:42 { name: NULL, }, May 16 23:25:42 { name: NULL, }, May 16 23:25:56 the processing will stop at the first null entry May 16 23:26:12 but with that you have a spare one for nvram backup May 16 23:27:34 indeed May 16 23:28:12 xMff, http://paste.ubuntu.com/608726/ May 16 23:28:16 that look more sane ? May 16 23:28:44 yes May 16 23:28:49 but group the 4 and 3 assignments May 16 23:28:52 do not mix May 16 23:29:46 oh and in if(get_router() ... May 16 23:29:53 two equal signs, not one May 16 23:30:28 I would also remove the startswith routine May 16 23:30:40 and use this instead: !strncmp(boardnum, "04FN", 4) May 16 23:30:54 kk May 16 23:31:00 * RealOpty needs to lean C May 16 23:31:28 the indentation in line 26 of the paste looks off too May 16 23:31:40 doublecheck tab/space etc. May 16 23:32:53 also behind the first NULL entry, put a comment like /* may be used for Simpleshare Backup NVRAM */ May 16 23:33:13 to prevent future editors to remove or regard it as accident May 16 23:33:19 kk May 16 23:36:00 xMff, http://paste.ubuntu.com/608730/ ???? May 16 23:36:42 yes May 16 23:37:00 kk time to finish making diner then test May 16 23:37:02 xMff, ty May 16 23:37:44 line 26, 27, 28 still need the whitespace lceaned May 16 23:37:48 cleaned May 16 23:40:23 I'll have to hit the bed. gn8 May 17 01:52:02 KanjiMonster: ping **** ENDING LOGGING AT Tue May 17 02:59:58 2011