**** BEGIN LOGGING AT Fri Dec 03 03:00:00 2010 Dec 03 03:16:39 mmmh. any wrt54g-vlan experts here? Dec 03 03:16:43 is vlan0 somehow magic? Dec 03 03:41:04 no Dec 03 04:54:24 heh. found my problem Dec 03 04:54:36 typo and uci didnt warn me ;) Dec 03 04:55:48 now the wrt has a soldered header Dec 03 08:11:41 acoul * r24221 /packages/libs/libiconv/patches/ (3 files): libs/libiconv: refresh patches Dec 03 08:16:34 acoul * r24222 /packages/libs/elfutils/ (4 files in 2 dirs): libs/elfutils: update to version 0.150 Dec 03 09:28:34 kaloz * r24223 /trunk/target/linux/mpc83xx/ (6 files in 4 dirs): [mpc83xx]: remove unused kernel versions, make 2.6.36 the default Dec 03 11:02:12 build #33 of adm5120 is complete: Failure [failed shell] Build details are at http://tksite.gotdns.org:8010/builders/adm5120/builds/33 Dec 03 11:05:07 ping thepeople Dec 03 11:06:42 jow_laptop: thanks for committing the iproute2 fix, this makes me happy :D Dec 03 11:26:58 DonDiego: np Dec 03 11:48:52 can someone please apply the openssl patch i submitted over a week ago to the ml? Dec 03 11:49:20 im using it here for over a week without any problems Dec 03 11:49:49 * drizztbsd needs to add classless support to udhcpc Dec 03 12:05:33 mirko * r24224 /trunk/package/libtool/ (5 files in 2 dirs): Dec 03 12:05:33 update libtool from version 1.5.26 to version 2.4, this includes libltdl (to v7) Dec 03 12:05:33 - WARNING: This switch is going to break something quite likely, however since Dec 03 12:05:33 more software is requiring libtool2/libltdl7 this change needs to be done some Dec 03 12:05:33 day anyway Dec 03 12:05:35 mirko * r24225 /packages/libs/gdbm/patches/ (. 100-fix-build-with-libtool2.patch): make gdbm using libtool2 the way it requires Dec 03 12:05:38 mirko * r24226 /packages/libs/libiconv/ (Makefile patches/200-work-with-libtool2.patch): [packages/iconv] fix linking issue when using libtool2 Dec 03 12:17:37 dingo * r24227 /packages/net/corkscrew/ (. Makefile): [patchteam] - Corkscrew - new package - Signed-off-by: koniu Dec 03 13:17:37 tripolar: pong Dec 03 13:19:09 thepeople hi, when will i get write access to tor and privoxy? Dec 03 13:19:14 i cant write by now to this dirs Dec 03 13:21:04 tripolar: when I hear from the server admins, same as before Dec 03 13:21:15 tripolar: I will let you know when it is live Dec 03 13:21:22 okay Dec 03 14:11:40 mirko: ping Dec 03 14:14:55 KanjiMonster: pong Dec 03 14:19:18 thepeople did u look at the openssl patches by now? Dec 03 14:54:53 acoul * r24228 /trunk/toolchain/uClibc/config-0.9.32/common: toolchain/uClibc: push an upstream fix Dec 03 14:55:44 uhm Dec 03 14:55:52 I have to push a patch for xterm-256color support Dec 03 16:11:40 loswillios: ping Dec 03 16:45:21 acoul * r24229 /trunk/toolchain/uClibc/ (2 files in 2 dirs): toolchain/uClibc: sync to latest uClibc-0.9.32 Dec 03 16:58:38 xMff: I have to test this http://openwrt.pastebin.ca/2009731 yet to double-check it, but what do you think? Dec 03 17:00:09 hmmm....just a sec I think I did the diff wrong Dec 03 17:01:03 cut -f1 -d\ >$er_md5sum_file 2>/dev/null Dec 03 17:01:04 why? Dec 03 17:01:20 cut returns error? Dec 03 17:04:17 xMff: http://openwrt.pastebin.ca/2009736 - I was appending intead of replacing when I output the diff Dec 03 17:05:03 drizztbsd: no dd always uses stderr Dec 03 17:05:15 drizztbsd: for the block copied count Dec 03 17:06:50 ping thepeople Dec 03 17:08:06 cshore: are you wrong Dec 03 17:08:20 + dd if=/dev/urandom count=32 bs=1k 2>/dev/null | md5sum | cut -f1 -d\ >$er_md5sum_file Dec 03 17:08:26 you have to do that :) Dec 03 17:08:38 stderr does not pass by pipe Dec 03 17:09:09 drizztbsd: it works, so I think you are mistaken Dec 03 17:09:38 cshore: he's right about the dd status output Dec 03 17:09:38 http://paste.pocoo.org/show/299590/ Dec 03 17:09:49 I'm not wrong :) Dec 03 17:10:26 drizztbsd: you're right....I don't know why I thought it worked Dec 03 17:10:36 np :) Dec 03 17:12:35 drizztbsd: probably confusing it with another script Dec 03 17:13:06 probably with () or something that stderr ignores Dec 03 17:13:57 drizztbsd: all I know is I have, in other scripts abused stderr and got it to work Dec 03 17:14:31 drizztbsd: but not through a pipe Dec 03 17:17:28 drizztbsd: anyway thanks...I'm muttering to myself trying to figure out what I had working; I can never remember the redirection rules Dec 03 17:20:48 cshore: ext4 works with mini_fo if you add "nobarrier" Dec 03 17:20:59 Kaloz: Jo told me Dec 03 17:21:29 Kaloz: but he also said there are reports the ext4 module with journalling is a problem for memory and resources Dec 03 17:21:54 Kaloz: you should use sync with nobarrier Dec 03 17:22:54 cshore: ? Dec 03 17:23:08 drizztbsd: so basically journalling is not really useful with nobarrier? Dec 03 17:23:31 nobarrier disables write barriers, so you can loose datas on power down Dec 03 17:23:36 Kaloz: some ppl. reported high load and ram usage when ext4 was mounted with journalling enabled Dec 03 17:23:37 Kaloz: You'll have to ask him for details Dec 03 17:23:38 drizztbsd: that's not true.. and well, you should sync anyways Dec 03 17:23:47 Kaloz: might have been a regression back then Dec 03 17:23:55 Kaloz: what is not true? Dec 03 17:24:42 drizztbsd: you "should" sync anyways, and it's not that dangerous imho Dec 03 17:25:04 it's dangerous if you does not sync and you use ordered as journal Dec 03 17:25:23 jow_laptop: if someone uses ext4 with journaling, the load will be as high as using ext4 with journaling. it's up to the user to pick what fs and why he/she wants it Dec 03 17:26:00 Kaloz: true. The point was more that it might be too demanding for a default setting Dec 03 17:26:05 Kaloz: You mean ext3 Dec 03 17:26:07 ? Dec 03 17:26:25 cshore: nah, I'm replying what jow_laptop said Dec 03 17:26:33 jow_laptop: dfault fs is ext4 without journal Dec 03 17:26:39 jow_laptop: instead of ext2 Dec 03 17:26:41 I know Dec 03 17:27:03 Kaloz: you ex4..as high as using ext4 with journalling Dec 03 17:27:09 Kaloz: so with extents and trees? Dec 03 17:27:19 cshore: yeah, I wanted to highlight that it won't increase load Dec 03 17:27:38 Kaloz: ah, ok.....it was probably a bug then Dec 03 17:28:04 Kaloz: there were issues with OOms and rapdid exhaustion of RAM Dec 03 17:28:10 *due to Dec 03 17:28:27 /dev/sda1 on /mnt/sda1 type ext4 (rw,relatime,barrier=1,data=ordered) Dec 03 17:28:30 that's by default Dec 03 17:29:19 using the automount feature Dec 03 17:29:57 what's the option for no journal? Dec 03 17:30:21 tune2fs Dec 03 17:31:30 drizztbsd: oh, I was hoping for something that didn't depend on another program Dec 03 17:32:06 drizztbsd: so it has to be chosen during mkfs or with tune2fs? Dec 03 17:33:13 i.e. we can't really do anything that will make the defaults no journalling? Dec 03 17:34:31 or how bad is journalling with nobarriers, compared to say ext3 with journalling? Dec 03 17:35:57 right now the default for extroot is ext3 with journalling Dec 03 17:38:22 mostly for historical reasons Dec 03 17:38:39 if we just do ext2 with sync that might be the best option for defaults Dec 03 17:39:07 if someone wants faster access they can enable the 'riskier' options on their own Dec 03 17:43:08 mirko: pong Dec 03 17:44:15 loswillios: committed all the changes needed by pulseaudio (libtool -> libtool2, gdbm-fixes, iconv-fixes) - will update the ticket later Dec 03 17:44:15 mirko: nice, I'll fire up a new build Dec 03 17:44:24 yeah, just saw it Dec 03 17:45:42 mirko: when does it segfault for you? Dec 03 17:46:12 when i start it :P Dec 03 17:46:13 xMff: what do you think for extroot fs defaults? Dec 03 17:46:37 they are fine except for ext4 Dec 03 17:46:43 that needs nobarrier too Dec 03 17:47:04 making an ext4 fs w/o journal is a doc item, no need to handle that on-target Dec 03 17:47:13 mirko: x86? Dec 03 17:47:18 loswillios: mips :P Dec 03 17:47:22 loswillios: i know, i know... Dec 03 17:47:26 jow_laptop: is it safe just to put nobarrier in the options, even for ext2, ext3? Dec 03 17:47:28 loswillios: should test it on i486 first Dec 03 17:47:42 ah Dec 03 17:47:52 jow_laptop: then we only need to doc the ext4 w/o journal Dec 03 17:47:57 cshore: don't know off-hand Dec 03 17:49:45 also, since we use sync by default, I think the default fs for extroot should be ext2 instead of ext3 ? Dec 03 17:50:11 no, ext4 Dec 03 17:50:15 oh, ok Dec 03 17:50:27 or at least ext3 Dec 03 17:50:34 but I see really no reason for 2 Dec 03 17:50:55 loswillios: do you have a target where it was running fine once? Dec 03 17:51:21 jow_laptop: was thinking because we're using sync anyway that journalling doesn't matter Dec 03 17:51:45 jow_laptop: and ext2 is the only one without journalling by default Dec 03 17:52:02 loswillios: just accepted the ticket (hope you don't mind) Dec 03 17:52:20 mirko: people reported it worked fine on brcm47xx Dec 03 17:52:23 sure Dec 03 17:53:03 loswillios: hmm, then it seems i did break it Dec 03 17:53:49 FWIW, it fails for me on x86 too. --help works, but --start bails out at some point Dec 03 17:54:03 and I haven't managed to debug that correctly yet Dec 03 17:56:09 loswillios: there's this guy in https://forum.openwrt.org/viewtopic.php?pid=104418 talking about appending "ac_cv_func_posix_fadvise=no" to CONFIGURE_ARGS helps with the segfault... Dec 03 17:56:17 did you use this option once? Dec 03 17:56:28 it's an hack Dec 03 17:56:56 someone should fix the configure.ac and sent the patch upstream (imho) Dec 03 17:58:01 mirko: no, I didn't get that far yet :p PA only throws a short segfault at me with no hint why it would fail at all. Dec 03 17:58:15 mirko: are you using uclibc-0.9.32? Dec 03 17:58:40 jow_laptop: are you still working on #6435? Dec 03 17:59:21 loswillios: yes, i am Dec 03 17:59:34 drizztbsd: no, in fact I forgot it Dec 03 17:59:42 doh Dec 03 17:59:43 mirko: me too and I'm really not sure if the bug's in PA or in uClibc. (note that the forum guy apparently didn't use 0.9.32, but somehting in backfire) Dec 03 17:59:52 my modem uses it and I had to do a bad hack Dec 03 17:59:57 bbl Dec 03 18:00:25 (bad hack means an hotplug script that does "route default dev #interface") Dec 03 18:00:47 well did you test those patches? Dec 03 18:00:59 not yet, I'm still at work Dec 03 18:01:12 loswillios: hmm, thought nptl is required anyway by PA - another guy sent a patch to the ticket, including GLIBC||EGLIBC||UCLIBC_0_9_32 within the DEPENDS-section Dec 03 18:01:31 btw I could port it to backfire Dec 03 18:01:38 using quilt or whatelse Dec 03 18:01:49 yes you could Dec 03 18:03:47 * drizztbsd needs to install openwrt under qemu for testing Dec 03 18:05:48 mirko: joined Dec 03 18:05:56 treffer: did you use uclibc 0.9.32? Dec 03 18:06:09 mirko: -ntpl, iirc Dec 03 18:06:31 mirko: I've pulseaudio running right now, preparing a full tarball in case anyone is interrested in config/makefile Dec 03 18:06:32 so, it was uclibc 0.9.3{0,1} (?) Dec 03 18:06:46 let me upload the config..... Dec 03 18:07:07 it was head after the last rc, build last weekend Dec 03 18:08:46 mirko: http://paste.openwrt.org/f6ac2a88 Dec 03 18:09:12 treffer: you said line 94-96 - what's about the other libs? they do not need to be in /usr/lib? Dec 03 18:09:34 mirko: exactly Dec 03 18:09:44 mirko: the *module* stuff is loaded by pulseaudio Dec 03 18:09:52 i see Dec 03 18:10:08 the depeding libs are loaded by the normal linker Dec 03 18:10:16 and this breaks.... with a segfault Dec 03 18:10:28 mirko: you could try a simple cp Dec 03 18:10:32 and see what happens Dec 03 18:10:35 on your router Dec 03 18:10:36 i see - how is that done on other systems (glibc, other distributions) Dec 03 18:11:04 why a "make install" of PA drops them there? Dec 03 18:11:18 and are you familiar what exactly "ac_cv_func_posix_fadvise=no" is doing? is that needed? Dec 03 18:11:23 well.... Dec 03 18:11:34 win 16 Dec 03 18:11:38 looks like other system mange to load them Dec 03 18:11:38 oops Dec 03 18:11:47 I've no idea if that's needed Dec 03 18:11:59 but there were reports about segfaults without it... in the forum Dec 03 18:12:12 yes, saw them Dec 03 18:12:23 however nobody said what that switch is actually doing Dec 03 18:13:01 well, when i arrive home i'm glad to checkout this out Dec 03 18:13:08 *check Dec 03 18:13:27 nice to hear that there is some action on the ticket Dec 03 18:13:51 i'd like to see it working upstream finally Dec 03 18:14:03 that would be awesome :-) Dec 03 18:14:21 nbd: i am about to leave, but i hoped to discuss the possible elimination of the use of fixed_phy Dec 03 18:14:22 should I upload my configs? Dec 03 18:14:35 Hey guys, fairly new to OpenWrt here... but I wrote a package for 6RD, that is very heavily based off of jow's 6in4 and 6to4 scripts. I also modified Luci's ifaces.lua to add support to it... where would I go to submit the new packaged and the modified source for review? "new ticket?" Dec 03 18:14:39 sn9: ok Dec 03 18:14:49 treffer: if it's just the libs i think it's not needed, however feel free.. Dec 03 18:15:11 mirko: well, avahi/dbus doesn't work as well as pulse, due to missing dbus config files..... Dec 03 18:15:42 methane: mailing list or ticket system, preferring the latter one :) Dec 03 18:15:51 Alrighty Dec 03 18:15:55 nbd: the problem is that fixed_phy refuses to work if another mdio bus is already registered, but adding the fixed_phy prevents phy drivers from loading Dec 03 18:15:59 treffer: have a simple fix here :) Dec 03 18:16:09 Gonna give it one final shakedown and review before I submit it Dec 03 18:16:14 mirko: ok :-) Dec 03 18:16:15 treffer: i don't wanna link pa against libavahi, requiring avahi-client Dec 03 18:16:22 had the same issue with mpd Dec 03 18:16:33 so... how do you announce your pulseaudio? Dec 03 18:16:48 I'm using it as a avahi-announced sink/server...... Dec 03 18:16:57 nbd: Wipster pointed out that internal phys are seen on a real mdio bus even without fixed phy, but just aren't driven Dec 03 18:18:04 if you see him online before i come back, maybe you can brainstorm Dec 03 18:18:24 treffer: https://dev.openwrt.org/browser/packages/sound/mpd/files/mpd.service?rev=21054 Dec 03 18:19:14 mirko: sounds reasonable Dec 03 18:19:46 treffer: so if avahi is installed, it's announced, if not, then not - so there's no hard-dependency on avahi Dec 03 18:19:54 nbd: ok? Dec 03 18:20:32 mirko: hmmm, but it means that a crashed pulseaudio won't get recognized, but that's probaply ok Dec 03 18:20:51 treffer: i think (at least for now) it is Dec 03 18:21:01 not perfect, but small (in size) and simple Dec 03 18:21:18 yeah, it's an embedded device, so size matters :-) Dec 03 18:21:41 yip Dec 03 18:23:06 sn9: sure Dec 03 18:23:10 thx Dec 03 18:25:44 loswillios: did you hear that? :) Dec 03 18:40:28 hi Dec 03 18:40:49 hi Dec 03 18:41:59 i have a question some time ago when i logged in into openwrt i could type thing while it still was loggin in with pub key authentication and this commands where executed when it login was successful - now everything i type in isnt execeuted while i type and the login is still processing - whats causing this? Dec 03 19:33:17 build #42 of ppc40x is complete: Failure [failed shell_4] Build details are at http://tksite.gotdns.org:8010/builders/ppc40x/builds/42 Dec 03 20:10:10 cshore * r24230 /trunk/package/block-extroot/ (3 files in 2 dirs): [package] block-extroot: Added md5sum to extroot and underlying root filesystem. This is used to ensure that the extroot is not mounted as the rootfs after a sysupgrade *until the user deals with the extroot). Dec 03 21:29:43 cshore * r24231 /branches/backfire/package/block-extroot/ (3 files in 2 dirs): Dec 03 21:29:43 [package] block-extroot: Merged 24230 from trunk: Added an md5sum to extroot and Dec 03 21:29:43 underlying root, which is used to ensure that extroot is not mounted after Dec 03 21:29:43 sysupgrade (until user deals with the necessary changes to the extroot) Dec 03 23:31:27 mirko * r24232 /trunk/package/libtool/Makefile: [package/libtool] add md5 checksum Dec 03 23:58:35 jow * r24233 /branches/backfire/target/linux/generic-2.6/ (4 files in 2 dirs): [backfire] merge r24195 Dec 04 00:20:00 mirko * r24234 /trunk/include/autotools.mk: [trunk/include/autotools.mk] disable libtool fixups in trunk globally Dec 04 01:40:56 nbd * r24235 /trunk/package/mac80211/ (35 files in 2 dirs): mac80211: update to wireless-testing 2010-11-30 Dec 04 01:41:03 nbd * r24236 /trunk/ (2 files in 2 dirs): ar71xx: work around a PCI controller bug which causes reads to the PCI_COMMAND register to return bogus values - properly fixes ath9k module reload issues Dec 04 02:49:30 build #34 of x86 is complete: Failure [failed shell_4 compile_6] Build details are at http://tksite.gotdns.org:8010/builders/x86/builds/34 **** ENDING LOGGING AT Sat Dec 04 02:59:59 2010