**** BEGIN LOGGING AT Wed Aug 18 02:59:57 2010 Aug 18 08:11:31 nico * r22693 /packages/net/openswan/ (4 files in 2 dirs): [packages] openswan: update to 2.6.28 (closes: #7544, #7725, #7782) Aug 18 09:22:49 [florian], ping Aug 18 09:22:57 <[florian]> Wipster: pong Aug 18 09:27:23 [florian], hope I'm not bugging you too much :) but I was thinking about this disabling internal phy and re-enabling, because there is a high and low port if one doesn't detect an external phy but does internal, and the other detect an external but not internal which one do you chose and how do you chose it if they are registered seperatly? Not sure if this case would exist as I duno if the 2nd interface is connected in some devices Aug 18 09:28:27 <[florian]> Wipster: well, most of the time, the second interface is not connected, but in some hardware designs it is Aug 18 09:28:47 <[florian]> Wipster: however, the MDIO bus, which allows controlling of PHYs and switches is shared between the two MACs Aug 18 09:34:26 [florian], hmmmm ok, say if you tested the connected interface first, disabled internal phy and enabled mii reading and it returned ok for the external PHY. Then probed the 2nd disconnected PHY and returned false on external and switched the chips internal back on and closed MII reading and failed again. Would that break the first? Aug 18 09:35:48 <[florian]> I hope not, but I am not sure how it would behave actually Aug 18 09:38:24 [florian], ok just a thought, my connected one is the low so it probes high properly with internal off and on and returns nout, then tries the 2ns and brings up the interface which is nice and I can see the real MII registers... the switch driver doesn't seem to start still and I'm not sure why... Aug 18 09:38:48 s/2ns/2nd Aug 18 09:42:48 [florian]: ping Aug 18 10:12:17 <[florian]> OutBackDingo: pong Aug 18 10:13:10 [florian]: u know anything bout the broadcom capabilities Aug 18 10:13:57 like do they support monitor mode and packet injection? Aug 18 10:14:12 <[florian]> OutBackDingo: not at all, sorry Aug 18 10:14:30 ok thanks Aug 18 10:19:10 <[florian]> OutBackDingo: I do think they support monitor mode and packet injection, but I have no idea how to do it Aug 18 10:38:29 <_trine> broadcom does do it now but not very well as far as I understand Aug 18 11:05:30 nico * r22694 /trunk/toolchain/gcc/ (Config.in Config.version): toolchain/gcc: remove remaining references to gcc-3.4.6 Aug 18 12:36:36 jow_laptop, lol someone opened a ticket when you did the kernel upgrade on wiki server. Aug 18 12:40:46 dingo * r22695 /packages/net/dudders/ (. Makefile files/ files/dudders.conf files/dudders.hotplug): Aug 18 12:40:46 [Patchteam] New Package Dudders - a utility for updating DNS records Aug 18 12:40:46 Signed-off-by: Dean Scarff Aug 18 13:58:44 mb * r22696 /trunk/ (Config.in include/package-defaults.mk include/package.mk): Support parallel build of certain packages Aug 18 14:04:21 mb * r22697 /packages/Xorg/lib/qt4/Makefile: qt4: Fix NPTL build options Aug 18 14:23:50 mb * r22698 /trunk/include/host-build.mk: Also support parallel build for host-packages Aug 18 14:51:54 is [mbm] still around? Aug 18 14:52:07 I haven't seen him talk or a commit in a LONG time Aug 18 15:26:26 mb * r22699 /packages/Xorg/lib/qt4/Makefile: qt4: Enable parallel build Aug 18 15:55:10 <[mbm]> Weedy: yeah, just not very active with openwrt these days Aug 18 15:58:29 zomfgbbq Aug 18 15:58:34 i thought you died Aug 18 16:01:13 juhosg * r22700 /trunk/target/linux/ar71xx/files/arch/mips/ar71xx/ (31 files): ar71xx: change mac address initialization Aug 18 16:01:17 juhosg * r22701 /trunk/target/linux/ar71xx/ (config-2.6.34 config-2.6.35): ar71xx: don't override CONFIG_SWAP Aug 18 16:01:19 juhosg * r22702 /trunk/target/linux/generic/files/drivers/net/phy/ (rtl8366_smi.h rtl8366rb.c rtl8366s.c): generic: rtl8366: add enable_vlan{,4k} to smi_ops Aug 18 16:01:21 juhosg * r22703 /trunk/target/linux/generic/files/drivers/net/phy/ (rtl8366_smi.c rtl8366_smi.h rtl8366rb.c rtl8366s.c): generic: rtl8366: add generic code to enable VLANs Aug 18 16:01:23 juhosg * r22704 /trunk/target/linux/generic/files/drivers/net/phy/ (rtl8366rb.c rtl8366s.c): generic: rtl8366: allow use of VIDs 16-4095 if vlan4k is enabled Aug 18 16:28:36 mb * r22705 /packages/Xorg/lib/qt4/Makefile: Aug 18 16:28:36 qt4: Don't build examples/demos, if they are not selected. Aug 18 16:28:36 This also adds PKG_CONFIG_DEPENDS to force a rebuild if those options change. Aug 18 16:28:36 This brings a significant build speed improvement. Aug 18 16:29:29 Ok, now I'm down from about 4 hours to 30 mins :) Aug 18 16:37:49 mb__: haha, you're down to 30 mins and you caused the next 10.03.1 build to take ~6 hours longer :P Aug 18 16:38:14 (every time the qt4 makefile is touched it makes the builds take an additional 6 hours) Aug 18 16:43:13 Bartman007: Well, that quickly outweights ;) Aug 18 16:43:51 Bartman007: I'm not doing that for fun, btw. I'm trying to reduce the pain before I implement and merge X11 support, which would have touchted the makefile anyway :) Aug 18 16:46:16 I know and I welcome the submake parallelization, it's something I had debated implementing for the same reason Aug 18 16:46:28 qt4 and freeswitch take far too long to compile Aug 18 16:50:21 hate bandwidthd Aug 18 16:50:41 Bartman007: I'm at about 50mins for complete qt4 on a quad PowerMac. Aug 18 16:51:58 you guys disabled postgres Aug 18 16:52:10 and i'm having a bitch of a time enabling it Aug 18 16:54:12 Weedy: make sure you set CONFIG_WEEDY=n Aug 18 16:55:54 lol Aug 18 16:59:46 [florian], ping Aug 18 17:03:26 Weedy: postgres is building for me... Aug 18 17:04:36 thats easy Aug 18 17:04:58 getting bandwidthd to pick it up and not have it segfault is the hard part Aug 18 17:09:15 <[florian]> Wipster: pong Aug 18 17:09:40 Weedy: oh, that makes far more sense. Aug 18 17:10:23 well I stopped the segfault Aug 18 17:10:32 now it just uses 100% cpu Aug 18 17:12:19 progress! Aug 18 17:16:44 [florian], been poking about today trying to find where the real switch PHY stopped being detected with the phy scan and I found it happened when it transitioned to kernel .30 at r17184 with the same cpmac driver code here is the output before and after. http://pastebin.com/TNqFaCZ4 Aug 18 17:18:08 <[florian]> Wipster: ok this is good to known, can you pinpoint a specific commit for this ? Aug 18 17:18:28 [florian], its interesting to see that disabling the internal phy in current trunk and doing the IO remap outputs the MII registers of PHY 0 (on the older kernel) with PHY 1F Aug 18 17:18:44 s/with/on Aug 18 17:20:24 [florian], well I have tested 17107 and 17186 thats my before and after, during that time 4 commits are made to the ar7 targ... got a feeling its probably the last commit, the one switching kernel versions from .27 to .30, maybe some mdio driver change duno Aug 18 17:21:32 <[florian]> maybe yes Aug 18 17:22:09 nbd: the current git version of relayd doesn't compile because of strict-aliasing in function add_arp ... Aug 18 17:23:06 nbd: don't you have the same issue? Aug 18 17:26:00 [florian], what would the next course of action be in tracking this missing phy down Aug 18 17:51:46 Bartman007: if you want parallel build for "freeswitch", you need to modify its makefile, btw. parallel is off by default. Aug 18 18:16:46 mirko: I'm merging the X11 stuff in a few simpler steps. That reduces chances to break something and also forces me to think about it again. Aug 18 18:18:01 I'm also going to change the way we select QWS vs X11. I think a "choice" option is the way to go here. Aug 18 18:18:37 This way it's also easier to add others (not sure if we'd need that. For xcb probably?) Aug 18 18:20:18 Anyway, I'm really trying hard to not introduce regressions. I'm test compiling since 10 hours or so :) Aug 18 19:03:29 mb * r22706 /packages/Xorg/lib/qt4/Makefile: Aug 18 19:03:29 qt4: Do not always compile optional features. Aug 18 19:03:29 This adds the packages to PKG_CONFIG_DEPENDS to make sure qt4 is rebuilt if the config changes. Aug 18 19:30:13 mb * r22707 /packages/Xorg/lib/qt4/patches/300-compile-fixes.patch: Aug 18 19:30:13 qt4: compile fix Aug 18 19:30:13 qRegisterAnimationInterpolator() is a function defined in a header. It is not declared static inline. Under certain circumstances this may cause linker errors. Make it static inline. Aug 18 19:52:18 Weedy: ? Aug 18 20:19:15 xMff: your just my goto guy for when i need to blame someone Aug 18 20:19:44 ah ok Aug 18 20:20:12 bandwidthd postgres support fffffffffffffffffffffffffffffffffffffffffffffff **** BEGIN LOGGING AT Wed Aug 18 20:55:40 2010 Aug 19 00:59:54 jow * r22708 /trunk/package/6in4/ (Makefile files/6in4.sh): [package] 6in4: bypass uci state when reading local IPv4 addr, specify TTL during tunnel creation (#7769) Aug 19 01:00:35 jow * r22709 /branches/backfire/package/6in4/ (Makefile files/6in4.sh): [backfire] merge r22708 Aug 19 01:06:57 jow * r22710 /trunk/package/6in4/files/6in4.sh: [package] 6in4: fix typo in last commit Aug 19 01:07:33 jow * r22711 /branches/backfire/package/6in4/files/6in4.sh: [backfire] merge r22710 Aug 19 02:44:10 FFFFFFFFFFFFFFFFFF Aug 19 02:44:21 bandwidthd keeps OOMing **** ENDING LOGGING AT Thu Aug 19 02:59:57 2010