**** BEGIN LOGGING AT Sun Nov 21 02:59:58 2010 Nov 21 03:30:14 build #28 of ixp4xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ixp4xx/builds/28 Nov 21 06:48:40 build #32 of s3c24xx is complete: Failure [failed shell_4] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/32 Nov 21 08:21:07 build #27 of etrax is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/27 Nov 21 08:53:00 build #36 of atheros is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/atheros/builds/36 Nov 21 10:58:30 build #34 of at91 is complete: Failure [failed shell] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/34 Nov 21 12:30:09 [florian]: ping Nov 21 14:00:44 jow * r24051 /trunk/package/dnsmasq/ (Makefile files/dnsmasq.init): (log message trimmed) Nov 21 14:00:44 [PATCH] Package dnsmasq init-file Nov 21 14:00:44 If an already running DHCP-server is found on Nov 21 14:00:44 the configured interface, the init-script refuses Nov 21 14:00:44 to start, but does not give a hint why. It took us Nov 21 14:00:45 some days to track down this behavior in our MANET, Nov 21 14:00:46 where every wifi/adhoc-node runs it own DHCP-server. Nov 21 14:01:23 jow * r24052 /branches/backfire/package/dnsmasq/ (Makefile files/dnsmasq.init): [backfire] merge r24051 Nov 21 14:13:44 <[florian]> KanjiMonster: pong Nov 21 14:16:38 hauke * r24053 /packages/libs/mysql/ (4 files in 3 dirs): mysql: update to version 5.1.53 Nov 21 14:19:17 hauke * r24054 /packages/libs/mysql/Makefile: mysql: decrease package number Nov 21 14:19:50 hauke * r24055 /packages/libs/openldap/ (4 files in 2 dirs): openldap: update to version 2.4.23 Nov 21 14:24:42 hauke * r24056 /packages/libs/postgresql/ (3 files in 2 dirs): postgresql: update to version 9.0.1 Nov 21 14:25:40 jow * r24057 /packages/libs/libxml2/Makefile: [packages] libxml2: enable debug module (#8289) Nov 21 14:28:57 jow * r24058 /packages/net/freeradius2/Makefile: [packages] freeradius2: Package rlm_always (#8291) Nov 21 14:32:05 hauke * r24059 /packages/net/freeradius2/ (3 files in 2 dirs): Nov 21 14:32:05 freeradius2: update to version 2.1.10 Nov 21 14:32:05 This closes #8252 and #7810 Nov 21 14:32:59 hauke * r24060 /packages/utils/coreutils/ (Makefile patches/001-no_docs_man_tests.patch): coreutils: update to version 8.7 Nov 21 14:48:08 hauke * r24061 /packages/utils/coreutils/patches/002-fix_compile_with_uclibc.patch: coreutils: add patch missing in r24060. Nov 21 15:07:38 build #34 of brcm47xx is complete: Failure [failed shell_13] Build details are at http://tksite.gotdns.org:8010/builders/brcm47xx/builds/34 Nov 21 15:20:00 build #36 of brcm63xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/36 Nov 21 16:13:16 If I wish to build an amd64 version of openwrt is the recommended way to build manually a cross compiler and then tell openwrt to use an external toolchain ? (cannot see any other way other than adding a complete new target) Nov 21 16:42:52 build #31 of ar71xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar71xx/builds/31 Nov 21 16:43:19 nbd, err, please pardon my bothering, but I guess the question is if you feel the IC+ IC175 port security "vlans" issue is to be handled with yet another swconfig config option or users are supposed to mess with mii's directly? Nov 21 16:50:43 nus: no idea Nov 21 16:51:21 heh, fair enough Nov 21 16:51:36 * nus ponders making a patch Nov 21 16:53:39 imo swconfig option would be cleaner as it limits the possible effects typos can cause ;) Nov 21 16:59:17 nbd * r24062 /trunk/package/mac80211/patches/520-ath9k_endian_fixes.patch: ath9k: fix some more endian issues Nov 21 16:59:22 nbd * r24063 /trunk/package/crda/Makefile: crda: update regulatory.bin based on latest wireless-regdb.git Nov 21 17:00:00 nbd * r24064 /branches/backfire/package/crda/Makefile: crda: merge regulatory.bin update from r24063 Nov 21 17:59:17 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/uscan.shtml Nov 21 18:16:01 ping KanjiMonster Nov 21 19:35:51 swalker: do you have a script to generate the table with the version numbers or do you do it by hand? Nov 21 20:34:15 nbd: ping Nov 21 20:34:21 pong Nov 21 20:34:33 can you commit my X86 patch? Nov 21 20:34:45 (please) Nov 21 20:41:57 is there an updated one that addresses kaloz' comments? Nov 21 20:59:41 I posted two versions... one which does what he suggests, and another that didn't. Either one gets the job done. Nov 21 21:00:11 got a link? Nov 21 21:55:23 build #27 of ifxmips is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ifxmips/builds/27 Nov 21 22:08:28 florian * r24065 /packages/net/yaddns/Makefile: [package] update yaddns to 0.1.2, thanks rhk Nov 21 22:08:31 florian * r24066 /packages/libs/libupnp/Makefile: (log message trimmed) Nov 21 22:08:31 [package] update libupnp to 1.6.8 Nov 21 22:08:31 this patch updates libupnp to 1.6.8. Nov 21 22:08:31 There are lots of fixes since 1.6.6, and 3 concerning embedded Nov 21 22:08:31 devices: Nov 21 22:08:31 - fix segfault when there is no services on embedded devices Nov 21 22:08:33 - customize the stack size of the threads used by pupnp Nov 21 22:08:34 florian * r24067 /packages/net/cups/Makefile: Nov 21 22:08:34 [package] update cups to 1.4.4 Nov 21 22:08:34 Update CUPS to current 1.4.4: Nov 21 22:08:34 "CUPS 1.4.4 fixes several security, scheduler, printing, and conformance Nov 21 22:08:35 issues." Nov 21 22:08:35 Also remove PKG_BUILD_DEPENDS, since zlib is already a normal dependency. Nov 21 22:08:36 Signed-off-by: Jonas Gorski Nov 21 22:08:36 florian * r24068 /packages/net/openvpn-devel/Makefile: (log message trimmed) Nov 21 22:08:37 [package] update openvpn-devel to week 45 snapshot Nov 21 22:09:11 [package] add fancontroled Nov 21 22:09:11 Add fancontroled (fancontrol for embedded devices) package. fancontroled Nov 21 22:09:11 is linux daemon written in C implementing a temperature dependent fan Nov 21 22:09:11 speed control. fancontroled provides similar functionalities like Nov 21 22:09:12 fancontrol script from lm-sensors; but fancontroled does not need bash Nov 21 22:09:12 or other dependencies (except lm-sensors library). That makes it Nov 21 22:10:58 build #29 of ps3 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/29 Nov 21 22:55:58 nbd: http://www.mail-archive.com/openwrt-devel@lists.openwrt.org/msg07277.html Nov 21 22:57:19 otherwise http://www.mail-archive.com/openwrt-devel@lists.openwrt.org/msg07276.html Nov 21 23:03:54 nbd * r24071 /trunk/target/linux/x86/ (geos/config-default net5501/config-default): Nov 21 23:03:54 x86: turn off CONFIG_X86_GENERIC for geos and net5501 Nov 21 23:03:54 CONFIG_X86_GENERIC obscures CONFIG_MGEODE_LX, etc. and causes it to use wrong cache-line size. Nov 21 23:03:54 Signed-off-by: Philip Prindeville Nov 21 23:04:04 i left out the part that touched the generic config Nov 21 23:05:42 ping xMff Nov 21 23:06:22 pong Nov 21 23:06:44 I just ran the iwinfo commands with the proprietary driver on 63xx. They appear to work Nov 21 23:07:05 cool Nov 21 23:07:07 (e.g. show wlan's I know exist for scan) Nov 21 23:07:39 the only one I haven't tested is assooclist with something associated to the device Nov 21 23:07:40 iwinfo wl0 freqlist / iwinfo wl0 txpowerlist and iwinfo wl0 countrylist produce meaningful results too? Nov 21 23:07:45 yes Nov 21 23:08:03 great Nov 21 23:08:47 how's the ui speed? Nov 21 23:09:14 I currently test on a fonera2, while its not exactly fast it is acceptable Nov 21 23:09:22 on chrome it's good Nov 21 23:10:32 nice Nov 21 23:10:56 hmmm.....prop driver crashes on enabling wifi Nov 21 23:14:43 grr, that sysupgrade breakage on trunk is really annoying Nov 21 23:16:51 nbd: what breakage? Nov 21 23:17:03 well, ldd segfaults Nov 21 23:17:17 and that breaks sysupgrade Nov 21 23:18:11 nbd: is the the busybox ldd? Nov 21 23:18:20 no, uclibc stuff Nov 21 23:18:33 nbd: ah ok, it was one or the other Nov 21 23:18:44 maybe we should just revert back to the old uclibc Nov 21 23:18:54 0.9.31 is so crappy Nov 21 23:19:17 Kaloz, what do you think? Nov 21 23:22:42 huh, ldd should be fixed by now Nov 21 23:22:51 since when? Nov 21 23:22:56 a week? Nov 21 23:23:03 for me it worked Nov 21 23:23:12 after applying some fixes to ldso Nov 21 23:23:17 i did do a dirclean recently Nov 21 23:23:23 maybe it wasn't recent enough Nov 21 23:23:23 hm ok Nov 21 23:23:40 but I agree, 0.9.31 is a trainwreck Nov 21 23:24:02 pthreads broken, resolvstuff (dnsmasq) broken, transmission broken due to random data corruption Nov 21 23:24:11 sysupgrade broken Nov 21 23:24:25 I think we should wait for 0.9.31.1 Nov 21 23:24:40 i think we should wait for 0.9.32 Nov 21 23:24:45 or that Nov 21 23:25:05 is there any advantage to using 0.9.31? Nov 21 23:25:14 did it fix or improve anything? Nov 21 23:25:20 yeah, it allows me to train my gdb skills -_- Nov 21 23:25:30 yes, it did improve my gdb skills *lol* Nov 21 23:26:06 sorry, had a clow fro breakfast. So far I only noticed regressions with .31 Nov 21 23:26:07 nbd: why? the generic code is wrong. that was part of the fix. Nov 21 23:27:04 maybe i misunderstood kaloz then? Nov 21 23:27:25 he mentioned something about changing the subtargets only and leaving the generic config alone Nov 21 23:27:41 and the targets that you changed did override the cache line size Nov 21 23:27:53 so which targets is the cache line size wrong for? Nov 21 23:28:28 the log2 of 64 is 6 the last time I looked. the file has 64 byte cache line size, but a shift of only 4 bits. Nov 21 23:30:52 so which targets is the cache line size wrong for? Nov 21 23:31:16 and by the way, the difference between the original patch and the alternate version of the patch (v2) is that v2 conformed to all of kaloz's constraints. Nov 21 23:31:51 after your patch, just target/linux/x86/generic/ is wrong. should be BITS=6 instead of BITS=4 Nov 21 23:32:11 i don't know which devices need which cache line sizes Nov 21 23:32:39 err, SHIFT=6 Nov 21 23:33:24 the i586 cache line size is (by default) 64 bytes thus a 6-bit shift. So the above file should contain CONFIG_X86_L1_CACHE_SHIFT=6 Nov 21 23:33:38 ok Nov 21 23:34:02 only specialized cores (like the Athalon and Geode) have a smaller cache line size. Nov 21 23:34:41 makes sense Nov 21 23:34:42 nbd * r24072 /trunk/target/linux/x86/config-default: x86: add missing l1 cache shift change Nov 22 00:04:47 xMff: btw. my build did have the ldd fix Nov 22 00:04:57 xMff: and the segfault shows up *after* it prints a few libraries Nov 22 00:05:19 yeah, at the point when ld.so wants to print itself Nov 22 00:05:47 how does the default linux version bump happen? is it done globally, or per platform, or what? Nov 22 00:05:53 per platform Nov 22 00:06:14 so what has to happen to bump the x86 stuff to 2.6.35.7 or later? Nov 22 00:06:35 nbd: I'd say lets scrap .31... Nov 22 00:06:38 build it, test it, submit a patch Nov 22 00:06:42 xMff: yeah Nov 22 00:06:55 i'm updating .32 to a recent git version as well Nov 22 00:06:59 we'll see how that works out Nov 22 00:07:09 meanwhile going back to 0.9.30 is a good idea Nov 22 00:07:24 maybe we can update the .30 micro version instead Nov 22 00:07:31 yeah Nov 22 00:07:32 .30.2 or what it was Nov 22 00:07:36 .3 i think Nov 22 00:08:52 Here's the proprietary driver oops on 63xx : http://openwrt.pastebin.com/StVJXUgY Nov 22 00:09:04 that looks ... useless Nov 22 00:09:13 how about turning on KALLSYMS? Nov 22 00:10:04 sure....just thought I'd post what I had before I rebuilt Nov 22 00:51:57 nbd * r24073 /trunk/toolchain/uClibc/ (9 files in 3 dirs): uClibc: update 0.9.32 to latest git Nov 22 00:52:43 I need to write a platform module for Geos that exposes 3 LEDS and 1 button, all GPIO based. what module can I borrow source from? Nov 22 00:59:04 xMff or nbd: any suggestions? Nov 22 01:34:03 build #31 of pxcab is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/31 Nov 22 01:41:05 grr, uclibc 0.9.32 also has ldd segfault issues Nov 22 01:49:50 well I don't think anybody did test this particular feature Nov 22 01:50:05 i'll try adding mark's fix and rebuilding Nov 22 01:54:27 looks better Nov 22 01:55:05 can you test whether dnsmasq works? Nov 22 01:55:19 in particular the --ptr-record option Nov 22 01:55:47 nbd * r24074 /trunk/toolchain/uClibc/patches-0.9.32/130-ldso-fix-__dl_parse_dynamic_info-segfault.patch: uClibc: 0.9.32 needs the ldso fix as well Nov 22 01:55:52 ok Nov 22 01:57:02 how do i test this? Nov 22 01:58:39 for example in /etc/dnsmasq.conf ptr-record=1.1.168.192.in-addr.arpa,test Nov 22 01:58:49 then nslookup 192.168.1.1 Nov 22 01:59:18 it should produce something meaningful Nov 22 01:59:43 yup Nov 22 01:59:44 that works Nov 22 02:00:06 for me it produces something like 192.168.2.10 user1.intern.domain..arpa which is completely wrong Nov 22 02:00:36 like a string corruption somewhere Nov 22 02:00:52 Name: 192.168.1.1 Nov 22 02:00:52 Address 1: 192.168.1.1 test Nov 22 02:00:59 ok, perfect Nov 22 02:01:39 maybe we should switch trunk to 0.9.32 then Nov 22 02:01:41 ;) Nov 22 02:02:16 transmission-daemon is another good candidate Nov 22 02:02:19 for testing Nov 22 02:03:57 marek * r24075 /packages/net/batman-advanced/ (3 files in 3 dirs): batman-adv: update package to latest release (2010.2.0) Nov 22 02:09:21 can you test this? Nov 22 02:09:48 not today anymore but I can look tomorrow Nov 22 02:10:00 ok Nov 22 02:10:39 is that ntpl stuff available for all our platforms now? Nov 22 02:10:46 not sure Nov 22 02:10:47 but i think so Nov 22 02:11:25 ok Nov 22 02:52:09 cool, with uclibc 0.9.32, iperf -s -i 1 works Nov 22 02:52:20 (that one never worked with any older uclibc version) **** ENDING LOGGING AT Mon Nov 22 02:59:58 2010