**** BEGIN LOGGING AT Thu Sep 16 02:59:57 2010 Sep 16 03:42:17 build #84 of rdc is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/rdc/builds/84 Sep 16 05:04:50 build #82 of s3c24xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/82 Sep 16 05:26:07 build #81 of ar7 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar7/builds/81 Sep 16 05:27:33 build #81 of ps3 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/81 Sep 16 05:33:35 buildslave2 needs git! Sep 16 05:44:16 build #114 of brcm63xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/114 Sep 16 05:44:19 build #101 of cobalt is complete: Failure [failed svn compile_12] Build details are at http://tksite.gotdns.org:8010/builders/cobalt/builds/101 Sep 16 06:07:22 build #86 of rb532 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/rb532/builds/86 Sep 16 06:07:41 build #91 of adm5120 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/adm5120/builds/91 Sep 16 08:50:47 <_trine> I have just flashed the newest trunk and now I cant connect to any websites on ipv4 Sep 16 08:51:46 <_trine> ipv6.google connects ok Sep 16 08:51:55 <_trine> but not other websites Sep 16 08:52:34 <_trine> I have tried restarting my networking and have even rebooted my computer Sep 16 09:17:29 build #84 of au1000 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/au1000/builds/84 Sep 16 09:18:20 is [florian] around here? Sep 16 09:26:52 <[florian]> dtatulea: I am yes Sep 16 09:26:59 <[florian]> dtatulea: still having issues with the malta qemu target? Sep 16 09:27:41 yep Sep 16 09:28:35 [florian]: thx for the response. Do you have any hints? Sep 16 09:28:51 <[florian]> dtatulea: can you remind me of the exact issue? it is after enabling pcnet32 right? Sep 16 09:32:10 [florian]: It's not related to pcnet32. It's the same if I enabled it or not: during init stage, it just doesn't do anything (it responds to key presses) Sep 16 09:32:37 <[florian]> dtatulea: I have no particular issues with the default configuratio Sep 16 09:33:45 Is it possible that stdio is forwarded somewhere else? Sep 16 09:34:05 <[florian]> I do not think so, unless you did not started qemu as shown in the readme, or you have a different qemu version than I have Sep 16 09:35:10 oh, yes. I forgot an important thing: I'm using the linux-mips tree, not the default... Sep 16 09:35:36 so it's not actually default Sep 16 09:37:00 <[florian]> so it's a regression in the linux-mips tree Sep 16 09:37:21 I will try with the default tree to see if it works Sep 16 09:38:19 [florian]: One more thing before I go. Did you use the qemu gdb server on the kernel? I'm using it with a vmlinux that has debugging symbols, but it has a strange behaviour Sep 16 09:38:41 continue doesn't work Sep 16 09:38:47 <[florian]> dtatulea: nope, just a plain Sep 16 09:38:49 <[florian]> kernel Sep 16 09:39:07 so you don't have experience with that, ok Sep 16 09:39:13 <[florian]> dtatulea: no sorry Sep 16 09:39:36 cool, thanks for the help. Checking the default kernel now. Sep 16 09:40:06 <[florian]> dtatulea: yes please do, that will definitively help you track down the issue, then you can git bisect linux-mips Sep 16 09:41:33 aye aye sir Sep 16 09:54:01 [florian]: Same thing with latest kernel: 2.6.35.4 Sep 16 09:54:12 s/latest/default Sep 16 09:58:24 <[florian]> dtatulea: strange, let me kick off a build Sep 16 09:59:08 wait, not just yet Sep 16 09:59:41 I'm not fully aquainted with the build system, so I might have built the wrong stuff Sep 16 09:59:45 it was too fast Sep 16 10:34:02 <_trine> is anyone around that can have a look at trunk Sep 16 10:34:12 <_trine> there is a problem with it Sep 16 10:34:33 <_trine> it seems not to be routing ipv4 Sep 16 10:44:17 * RealOpty is testing x86 for firewall issues. Sep 16 10:55:58 _trine, Sep 16 10:56:36 i dont have your device but im gonna try out x86 virtualbox. Sep 16 10:56:47 there were recent changes to firewall stuff. Sep 16 11:04:09 odd... i get a build error lol Sep 16 11:06:23 RealOpty: details? Sep 16 11:06:49 indeed. doing make V=99 now Sep 16 11:08:27 <_trine> PING 66.102.7.104 (66.102.7.104) 56(84) bytes of data. Sep 16 11:08:27 <_trine> ^C Sep 16 11:08:27 <_trine> --- 66.102.7.104 ping statistics --- Sep 16 11:08:27 <_trine> 117 packets transmitted, 0 received, 100% packet loss, time 116016ms Sep 16 11:09:11 _trine: iptables -nvL; iptables -t nat -nvL Sep 16 11:10:03 xMff, i think it failed at package/base-files Sep 16 11:10:19 find out when it gets to it :/ Sep 16 11:10:28 <_trine> I cant access pastebin Sep 16 11:10:35 <_trine> shall i paste it here Sep 16 11:10:38 no Sep 16 11:10:42 lol Sep 16 11:10:51 <_trine> its not that big Sep 16 11:11:14 http://ipv6.pastebin.com/ Sep 16 11:12:19 http://iloveopenwrt.pastebin.com/ try that one Sep 16 11:12:30 ah right :P Sep 16 11:13:26 xMff, http://iloveopenwrt.pastebin.com/22VB87ak Sep 16 11:13:27 pastebin.ca is v6 though Sep 16 11:13:53 idk why its not finding base-files ? Sep 16 11:13:56 <_trine> they wont open Sep 16 11:14:04 <_trine> shall I send it you in pm Sep 16 11:14:07 RealOpty: make defconfig Sep 16 11:14:15 mh, the 6in4-scripts work surprisingly well in my r15k kamikaze Sep 16 11:14:50 _trine: yes pm is okay Sep 16 11:15:37 RealOpty: grep ARCH_PACKAGES .config Sep 16 11:15:51 xMff, i had a config that i used just a few days ago Sep 16 11:15:52 ok Sep 16 11:16:08 _trine: your firewall is empty Sep 16 11:16:18 xMff, nothing returned. hmm Sep 16 11:16:27 RealOpty: after make defconfig ? Sep 16 11:16:46 nope. doing that now :) Sep 16 11:16:56 <_trine> my firewall looks normal Sep 16 11:17:17 xMff, CONFIG_TARGET_ARCH_PACKAGES="x86" Sep 16 11:17:27 RealOpty: make should work now Sep 16 11:17:45 ah ok ty :) Sep 16 11:17:51 do i need to re edit config? Sep 16 11:17:52 <_trine> so it looks likes its a firewall issue Sep 16 11:18:08 _trine: yeah, the issue is that the firewall isn't running Sep 16 11:18:25 _trine: FW_TRACE=1 fw reload Sep 16 11:18:45 nope didnt have to Sep 16 11:18:47 thanks again. Sep 16 11:19:40 ha i said firewall!!! :) Sep 16 11:20:24 <_trine> xMff, there is a lot of this Sep 16 11:20:34 <_trine> do you still want it in pm Sep 16 11:20:39 _trine: no Sep 16 11:20:48 just check whether ipv4 works afterwards# Sep 16 11:20:56 yeah use pastebin :) Sep 16 11:21:14 j/k :) Sep 16 11:21:23 <_trine> no its just the same Sep 16 11:21:41 iptables -nvL still identically? Sep 16 11:22:50 <_trine> xMff, its not the same,, its a bit less now Sep 16 11:23:04 thats odd Sep 16 11:23:42 _trine: please run this: iptables -t nat -A POSTROUTING -o $(uci get network.wan.ifname) -j MASQUERADE Sep 16 11:24:03 its still empty Sep 16 11:24:30 <_trine> still empty Sep 16 11:24:44 <_trine> oh Sep 16 11:24:50 <_trine> its working now Sep 16 11:25:08 workarounded, not fixed Sep 16 11:25:38 now: FW_TRACE=1 fw reload Sep 16 11:25:45 then: iptables -t nat -A POSTROUTING -o $(uci get network.wan.ifname) -j MASQUERADE Sep 16 11:25:51 finally: pastebin Sep 16 11:26:21 build worked. Sep 16 11:29:41 _trine: ahh Sep 16 11:29:54 I knew the fw_die in that particular place was a bad idea Sep 16 11:30:32 _trine: in yuer /etc/config/firewall, how does the "CentOSssh" redirect looks like? Sep 16 11:30:42 <_trine> xMff, I'm glad I have serial :) Sep 16 11:32:01 <_trine> xMff, I noticed that when I restarted the firewall it didnt like that redirect but its the first time I have ever seen it complain Sep 16 11:32:15 _trine: yeah the complaining is new Sep 16 11:32:38 the guy who wrote firewall 2 initially made some errors fatal that where jsut warnings in the old fw Sep 16 11:32:46 a fatal error will abort and rollback Sep 16 11:32:58 I'll change that to a warning + rule skip Sep 16 11:33:13 however I'd like to see the redirect, how its configured Sep 16 11:33:21 <_trine> http://pastebin.com/vTqB90Ry Sep 16 11:33:39 _trine: ok change the target to DNAT Sep 16 11:33:56 or remove the target option entirely Sep 16 11:34:04 then reload Sep 16 11:34:14 <_trine> I have a few similar to that Sep 16 11:34:31 target ACCEPT isn't valid for redirects Sep 16 11:34:45 <_trine> ok I will change them Sep 16 11:35:02 previously it was just ignored, now that this option is defined for redirect section it must have a valid value Sep 16 11:38:34 nbd * r23078 /trunk/package/mac80211/patches/601-rt2x00-fix-hang-on-ifdown.patch: rt2800pci: add back ifdown hang workaround that was accidentally dropped during a compat-wireless update Sep 16 11:39:44 note to self: don't mount the the squashfs rootfs as jffs2 next time, at least not as rw Sep 16 11:40:02 :P Sep 16 11:40:08 hi KanjiMonster Sep 16 11:40:11 nbd * r23079 /branches/backfire/package/mac80211/patches/601-rt2x00-fix-hang-on-ifdown.patch: rt2800pci: add back ifdown hang workaround from r23078 Sep 16 11:40:20 <_trine> xMff, I changed the redirects and restarted the firewall and its working now Sep 16 11:40:45 hi stintel Sep 16 11:40:50 <_trine> shall I try a reboot Sep 16 11:41:00 _trine: no need to Sep 16 11:41:36 <_trine> well I will have to reboot anyway so I can put my 2 torx screws back :P Sep 16 11:41:50 <_trine> I threw the other 2 away Sep 16 11:41:53 <_trine> :) Sep 16 11:42:54 <_trine> thanks xMff Sep 16 11:47:42 jow * r23080 /trunk/package/firewall/ (4 files in 2 dirs): [package] firewall: make invalid redirects and duplicate zones non-fatal, print a notice and discard them Sep 16 11:55:05 what is the need to declare config files on the package makefile? Sep 16 11:55:18 define Package/conffiles Sep 16 11:55:18 ... Sep 16 11:55:18 endef Sep 16 11:55:28 are that to list what needs to be preserved on firmware update? Sep 16 11:55:33 yes, but why? Sep 16 11:55:34 no Sep 16 11:55:50 why? because opkg can't figure that out by itself Sep 16 11:56:08 why aren't they regular application files... Sep 16 11:56:14 ? Sep 16 11:56:26 config files get special treatment Sep 16 11:56:45 they're not overwritten for example if a package gets upgrade and the original conf was modified Sep 16 11:57:50 sysupgrade x.bin will preserve that files or only the ones at /etc/config/*? Sep 16 12:00:28 currently only the ones in /etc/config Sep 16 12:00:34 plus some assorted in /etc Sep 16 12:04:02 <_trine> everything is back together again now and its all working fine ,, thanks all Sep 16 12:07:15 great Sep 16 12:07:38 I think I will attempt to migrate my WDS to the WNDR3700 today Sep 16 12:10:02 <_trine> I'm glad you have a wndr3700 :) Sep 16 12:10:54 its damns fast Sep 16 12:11:11 firewall reload takes a second or so while its like 20 seconds on a fonera2 Sep 16 12:11:31 <_trine> yes its a big jump from a fonera Sep 16 12:18:38 xMff: :) Sep 16 12:25:30 anyone any experience with oprofile? Sep 16 12:27:28 I'm trying to find out why rootfs on usb results in slow writes on ar71xx Sep 16 13:26:18 build #82 of ramips is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ramips/builds/82 Sep 16 14:11:41 [florian]: Ok, after make clean && rm .config it works Sep 16 14:12:16 s/make clean && rm .config/make clean && rm .config && make Sep 16 14:13:19 dtatulea: case closed then? Sep 16 14:14:49 I guess. I will try to compile the linux-mips kernel to see what happens. If it works, I will enable some stuff to find the culprit. Sep 16 14:14:55 Thanks a lot for the help! Sep 16 14:15:27 great Sep 16 15:12:32 cshore: the users experiencing dns troubles with extroot reported that they fixed it by formatting the drive first, guess they ran on top of fat32 or something Sep 16 15:14:09 ah, I should have thought of that Sep 16 15:14:48 it works too well as it seems ;) Sep 16 15:15:00 heh Sep 16 15:48:08 cshore: tried switching to 2.6.35, didn't change a thing Sep 16 15:48:56 cshore: now I compiled a debug kernel, trying to see if I find anything through oprofile Sep 16 15:50:07 ok...I hope so because I admit I have no idea where to start looking - I don't see why it should matter that is a pivoted root vs regularly mounted filesystem Sep 16 15:50:28 it's not the overlay, which I thought it might be Sep 16 15:50:57 so it must be something to do with the pivot (is all I can think of) Sep 16 15:52:40 it this with sync write or not synced writes? and reading is slow or only writing? Sep 16 15:52:53 not-synced=cached Sep 16 16:02:39 cshore: m(. I should have noticed that muuuch earlier. it's the "sync" that's making it that slow; my reference was always mount with "mount" (which didn't make it that slow). Removed the sync and the speed is now fine Sep 16 16:03:40 ... Sep 16 16:03:40 although I'm wondering why it's that slow ... the same test with && sync is still magnitudes faster on not-synced than synced Sep 16 16:03:55 maybe unsuitable block size Sep 16 16:04:04 ah, that might be it Sep 16 16:04:24 do you really need snc? Sep 16 16:04:27 sync Sep 16 16:05:13 this was to make sure it isn't keeping stuff in the buffer but actually finished writing it Sep 16 16:06:19 will it basically bypasses caching Sep 16 16:06:43 and the kernel cannot sync in background / io ops are blocked until the cache got synced Sep 16 16:06:59 s/io/write/ Sep 16 16:12:53 yes, that was my intention (else I won't get anywhere near the real write speed to the device) Sep 16 16:14:34 well, that tought me to not blindly accept default values ;) Sep 16 16:17:38 cshore: sorry for the false alarm, and thanks for the final hint Sep 16 16:18:09 KanjiMonister: np, I'm glad we figured it out Sep 16 16:18:14 sounds like the wiki needs some troubleshooting points on this topic Sep 16 16:19:04 xMff: yes Sep 16 16:19:31 xMff: extroot does seem to have a number of things people have issues with Sep 16 16:23:12 xMff: almost need a mini-FAQ just for it Sep 16 16:56:55 xMff: to sshfs mount on files on an openwrt router what package(s) does the openwrt box need to have installed? Sep 16 16:57:09 -the first on Sep 16 16:57:27 mount openwrt on a pc or mount a pc on openwrt? Sep 16 16:57:34 openwrt on a pc Sep 16 16:57:43 openssh-sftp-server Sep 16 16:57:50 ok, thanks Sep 16 16:57:51 it works with dropbear Sep 16 16:58:34 cool....there another package that is supposed to be an sftp replacement in the repository too, isn't there? Sep 16 16:58:47 don't know about one Sep 16 16:58:54 http://stackoverflow.com/questions/184618/what-is-the-best-comment-in-source-code-you-have-ever-encountered/549611#549611 :) Sep 16 17:04:22 I like "O(scary)" Sep 16 17:04:49 :) Sep 16 17:13:09 cshore: btw, does block-extroot even work if its not included in the image but intalled later through opkg, especially if the required kernel modules were installed later? (this is regarding https://dev.openwrt.org/ticket/7949 ) Sep 16 17:13:25 KanjiMonster: no Sep 16 17:13:34 thought so Sep 16 17:13:41 well Sep 16 17:13:43 KanjiMonster: only using a pure jffs2 filesystem Sep 16 17:13:45 no on squashfs images Sep 16 17:13:55 yes on ext, jffs, ubi Sep 16 17:16:13 well mtdblock3 as overlay => sqashfs-image; if it were an jffs2-image it would be mtdblock2 Sep 16 17:16:27 right Sep 16 17:16:41 so last comment is right on the spot Sep 16 18:04:15 With 2.4 gone is it impossible for me to get my patches against backfire included for new router that is kernel2.4? Sep 16 18:04:37 :( Sep 16 18:10:30 well, you can port stuff to 2.6 :) Sep 16 18:12:33 i only got serial to work. Sep 16 18:13:11 well that's not true, i got everything working except the ethernet ;) Sep 16 18:15:18 their tulip driver hack just basically hardcodes in the resources since they're not visible on pci... and i basically did that in the 2.6 tulip driver, and for whatever reason it doesn't work the same. Sep 16 19:18:36 jow: what do you think of the next new version (i.e. not backfire but the next full release) including extroot in the builds by default? Sep 16 19:18:47 oh nvm Sep 16 19:18:54 that would require usb packages Sep 16 19:19:09 not just the extroot packages Sep 16 19:19:44 trying to think how to make extroot available without users having to compile their own image Sep 16 19:20:59 maybe we can include the preinit part of it Sep 16 19:21:08 and only leave the rest in extroot Sep 16 19:26:04 you mean modify extroot to load the modules from the jffs2? Sep 16 19:27:31 is there some big technical hurdle I don't see that would prevent jffs2 mounting and finally a switch over? Sep 16 19:28:10 you could also extend preinit to re-scan for parts in /overlay/lib/preinit/ once its beyound the mount jffs2 stage Sep 16 19:28:19 Well, if I modify it so that will load the (usb say) modules from the jffs2 then it would work Sep 16 19:28:39 right Sep 16 19:29:32 /tmp/overlay, but yeah Sep 16 19:29:56 that way extroot doesn't even need to be in the image Sep 16 19:32:40 basically the preinit_main hook could be split into before mount jffs2 and after, with after being included the results scanning /tmp/overlay/lib/preinit Sep 16 19:33:35 just introduce preinit_overlay Sep 16 19:33:37 s/being included/including the the results Sep 16 19:33:39 How does one reproduce build environment for e.g. 10.03.1-rc3? I searched wiki to no avail :( Sep 16 19:33:49 xMff: or that Sep 16 19:33:55 :) Sep 16 19:34:32 I was thinking in terms of being able to override the preinit_main, but that can already be done by overriding functions Sep 16 19:35:46 dottedmag: its r22796 of branches/backfire/ with default package set + "luci" Sep 16 19:36:34 xMff: is there config published? Or how to recreate it? Sep 16 19:36:49 dottedmag: so echo CONFIG_TARGET_$arch=y > .config; echo CONFIG_PACKAGE_luci=y >> .config; ./scripts/feeds update; ./scripts/feeds install -a; make defconfig; make Sep 16 19:36:56 xMff: ah, thanks Sep 16 19:37:16 CONFIG_ALL=y too Sep 16 19:37:36 xMff: I also want to work on sysupgrade so that the stuff in the luci config is in system config instead Sep 16 19:37:48 xMff: as well as working probably for extroot Sep 16 19:38:54 dottedmag: http://kamikaze.openwrt.org/8.09.1/release-build.sh Sep 16 19:39:12 build #104 of octeon is complete: Failure [failed compile_7] Build details are at http://tksite.gotdns.org:8010/builders/octeon/builds/104 Sep 16 19:42:41 meh, why do I always notice *after* flashing I forgot to apply a local patch Sep 16 19:48:22 also, I'm think we should remove the busybox blkid and swap-utils and use util-linux. That's because they add to size when they're not needed, plus the busybox blkid lacks the ability to detect volume type, which is a useful capability for block-hotplug Sep 16 19:49:18 e Sep 16 19:49:21 whoops Sep 16 19:50:29 s/think/thinking/ Sep 16 20:01:28 an one last thing I'm thinking of (I need to do this for extroot at the least) is to add the ability to, on first boot, contact the packages server and opkg install a list of packages, which can be either predefined in menuconfig, from a list in /etc/somewhere, or downloaded from the server. Sep 16 20:01:28 Do you think this could be useful to trim the list of packages included in the default images as well, or is the loss of space due to jffs2 size too much of a problem? An interesting thing for this would be to have a cgi that takes a prepared image, except the trx or imagetag or whatever header and adds that, and inserts a URL for the list of packages to download, so that much more customized images could be created Sep 16 20:02:31 That last would be a lot of work, but I think it'd be fun to do Sep 16 20:02:45 kind of the rom-o-matic for gpxe and etherboot Sep 16 20:02:51 *like Sep 16 20:23:04 Is 'network' option of wifi-interface config logically means adding wireless interface to ifname option (and hence works only for bridges or interfaces with empty 'ifname' options)? Sep 16 23:24:55 cshore, speaking of util-linux-ng. why are we using a really old build? Sep 17 00:29:19 build #116 of ar71xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar71xx/builds/116 Sep 17 00:37:41 dottedmag: yes, correct Sep 17 00:53:28 build #99 of gemini is complete: Failure [failed shell_3] Build details are at http://tksite.gotdns.org:8010/builders/gemini/builds/99 Sep 17 01:46:53 ping cshore Sep 17 01:49:56 xMff, cant find that link you showed me yesterday with all the hardware listedc. Sep 17 01:53:14 nvm Sep 17 01:53:15 http://wiki.openwrt.org/oldwiki/openwrtdocs/hardware Sep 17 01:57:32 hm the block-* package family needs some changes imo Sep 17 01:57:46 block-mount and block-hotplug should be merged Sep 17 01:57:55 block-extroot should be in base system too Sep 17 01:58:07 it also should select usb storage instead of depending on it Sep 17 01:58:16 and it should enable filesystems as well Sep 17 02:22:30 ^ Sep 17 02:22:56 apart from that it works well Sep 17 02:23:24 ^ Sep 17 02:59:30 ping xMff **** ENDING LOGGING AT Fri Sep 17 02:59:57 2010