**** BEGIN LOGGING AT Mon Aug 16 02:59:56 2010 Aug 16 06:05:31 acoul * r22666 /trunk/package/busybox/patches/920-macosx-endian.patch: package/busybox: fix macosx cross compile, closes #7775 Aug 16 09:13:14 nico * r22667 /trunk/package/grub/Makefile: package/grub: add a prereq check for 32 bits host development files when building on x86_64 (closes: #7753) Aug 16 09:45:49 lars * r22668 /trunk/include/package.mk: Aug 16 09:45:49 Add build variants support to InstallDev Aug 16 09:45:49 Before building a build variant would have removed the staging files installed Aug 16 09:45:49 by the previous build variant from the same makefile. Aug 16 09:49:11 lars * r22669 /trunk/package/ncurses/Makefile: Aug 16 09:49:11 [package] ncurses: Add libncursesw package Aug 16 09:49:11 Thanks bartbes Aug 16 09:58:14 nico * r22670 /branches/backfire/package/grub/Makefile: Aug 16 09:58:14 [backfire] merge r22667 Aug 16 09:58:14 package/grub: add a prereq check for 32 bits host development files when building on x86_64 (closes: #7753) Aug 16 10:09:39 nico * r22671 /packages/utils/owfs/Makefile: Aug 16 10:09:39 [packages] owfs: sanitize Makefile: Aug 16 10:09:39 * use Build/Configure/Defaults and CONFIGURE_{ARGS,VARS} Aug 16 10:09:39 * add owget binary to owshell package, install shared libs dev symlinks (closes: #7749) Aug 16 10:09:39 * use coherent template names Aug 16 10:09:40 * remove obsolete UninstallDev Aug 16 10:09:40 * bump release number Aug 16 10:15:11 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/uscan.shtml Aug 16 10:56:30 jow_laptop, ping Aug 16 11:01:35 here Aug 16 11:08:42 jow_laptop, should the option chrooteveryone 1 be in the default pure-ftpd config file? Aug 16 11:30:02 no idea Aug 16 11:30:03 maybe Aug 16 12:05:13 jow_laptop, /sys/class/leds <- shouldnt my leds be listed here too? Aug 16 12:05:25 they are listed in /proc/diag/leds Aug 16 12:05:58 im asking cause http://wiki.openwrt.org/doc/uci/system#leds says to look at that location Aug 16 12:06:09 i think that needs to be edited to /proc/diag Aug 16 12:06:19 /leds Aug 16 12:06:53 but i think that /proc/diag/leds is only for brcm47xx. so maybe /sys/class/leds is broken ? Aug 16 12:31:27 more likely its jsut not supported on brcm47xx Aug 16 12:36:55 jow_laptop, im updating wiki. is 10.03.1-rc1 the current 'stable' branch? Aug 16 12:38:47 no Aug 16 12:38:55 its a release candidate Aug 16 12:39:01 current stable is 10.03 Aug 16 12:39:09 kk making sure Aug 16 12:43:09 jow_laptop: was it cut from a recent trunk ? Aug 16 12:43:18 no Aug 16 12:43:23 from backfire branch Aug 16 12:43:29 i think its a ^^ yeah lol Aug 16 12:43:42 which is basically 10.01 with all the upstream patches Aug 16 14:16:36 Hauke1, ping Aug 16 14:28:53 [florian], ping Aug 16 14:29:35 <[florian]> Wipster: pong Aug 16 14:33:33 [florian], hey got my serial attached finally. I found that if I take off the fixed PHY stuff in platform.c the router doesn't bring up any interface. I reverted the fixed phy stuff in CPMAC back to char * for mdio bus and it comes up with duplicate mdio bus/0, this is with trunk.. I had success bringing the ethernet back up at kernel .27 and I saw the correct registers. Aug 16 14:33:53 anyone know the commands to make a package im trying to make appear in makemenu? Aug 16 14:34:02 or just make them into the .ipk file? Aug 16 14:34:18 <[florian]> Wipster: okay, I do not quite understand what you modified actually Aug 16 14:34:37 <[florian]> Wipster: if you want to get rid of the fixed PHY registration, all of the code is located in arch/mips/ar7/platform.c Aug 16 14:35:10 <[florian]> Wipster: about the MDIO bus registration, the name is just there to create a unique object for the kernel Aug 16 14:36:38 [florian], ahhhh in which case just reverting fixed PHY in platform.c ends up with no eth interfaces being detected Aug 16 14:38:18 <[florian]> Wipster: yes, because the PHY detection logic did not manage to detect any PHY Aug 16 14:38:24 RealOpty: pong Aug 16 14:38:33 <[florian]> Wipster: I will cook you a patch when I am back home for you to test with Aug 16 14:41:07 [florian], ok thanks I appreciate it, still getting my head around it got mislead by a patch named cpmac_fixed_phy which was tweaking mdio_bus_is's :) Aug 16 14:41:28 <[florian]> Wipster: the tweaking of the mdio_bus_is is just cosmetic actually ;) Aug 16 14:42:23 <[florian]> Wipster: the real trick is to play with the external PHY enable/disable thing Aug 16 14:43:25 [florian], ahhh makes a bit more sense now, where about is that? I didn't spot it... saw external_switch Aug 16 14:44:36 <[florian]> Wipster: in the case of external_switch being set, it tries to use the platform supplied phy_id Aug 16 14:44:42 <[florian]> Wipster: which, if not unset defaults to 0 Aug 16 14:44:47 <[florian]> Wipster: which might actually be a sane setting Aug 16 14:45:58 <[florian]> Wipster: also the probing logic should be rewritten Aug 16 14:47:51 <[florian]> Wipster: at least the mii probing could be moved to its own function Aug 16 14:49:50 [florian], fair enough. In the times before the fixed PHY in platform.c how did it detect a PHY to bring up the interfaces? Was it just because in adjust_link priv->phy->link was assumed to be always true or more to it? Aug 16 14:50:49 <[florian]> Wipster: well, when the connection between CPMAC and the PHY works, CPMAC is able to read the PHY status register and check for the link bits Aug 16 14:50:58 <[florian]> Wipster: this is how it "knows" if the link is still up or not Aug 16 14:52:03 [florian], so thats why before fixed phy it reported link is down... yet worked because the stack thought all was well? Aug 16 14:52:55 <[florian]> Wipster: it could not read the phy status register, but we tricked to continue sending packet despite the link status not being read correctly Aug 16 14:53:43 [florian], I see Aug 16 14:58:20 [florian], so is the PHY status register in the switch, or in the ar7 or indeed both. Aug 16 14:58:46 <[florian]> Wipster: well, depending on the kind of switch, the "usual" method for reading the link status may not work Aug 16 14:58:55 <[florian]> Wipster: because a switch usually has a different register layout than a PHY chip Aug 16 14:59:52 <[florian]> Wipster: so the content of the register read at the usual location are most likely to be incorrect Aug 16 15:02:01 [florian], ok does the ar7 have a PHY interface on it and it just operates in MAC? mode when there is a switch? Aug 16 15:02:15 <[florian]> Wipster: the switch is external Aug 16 15:02:29 <[florian]> Wipster: if I recall right, AR7 has an internal PHY on one of the two cpmacs Aug 16 15:02:36 <[florian]> Wipster: but can also be disabled to use an external PHY Aug 16 15:03:56 [florian], is there anything which uses the internal PHY, I guess routers would normally have a switch chip. Aug 16 15:06:35 <[florian]> Wipster: yes, but the default might be to use the internal PHY actually Aug 16 15:06:59 <[florian]> Wipster: so we should be more intelligent and check if the internal PHY is enabled or not, disable it, try to probe for the switch, if unsuccessful, re-enable it Aug 16 15:11:36 [florian], oh right yeh that sounds better, I did see a patch for disabling internal PHY from narge. So would the general flow be disable internal phy, probe MII and see if a switch driver registers, ask the switch for the phy status location (or the register contents). If it fails re-enable internal phy and register a fixed? Aug 16 15:12:25 <[florian]> Wipster: something like that yes Aug 16 15:12:37 <[florian]> Wipster: we now have enough switch drivers being able to detect them successfully Aug 16 15:16:57 [florian], nice, well thanks for your time explaining what goes on. Aug 16 15:18:10 <[florian]> Wipster: no problem, I also want this to be resolved ;) Aug 16 15:18:37 [florian], :) Aug 16 15:22:51 <_trine> thepeople, ping Aug 16 15:58:06 [florian], just a quick one... how did you trick it to bringing up eth0 and 1 before fixed PHY? Aug 16 16:01:21 florian * r22672 /trunk/package/acx-mac80211/Makefile: [package]: update acx-mac80211 and change the git url. Aug 16 16:08:32 nico * r22673 /branches/backfire/target/linux/x86/image/Config.in: [backfire] x86/image: restore grub console for generic images (closes: #7712) Aug 16 16:12:01 [florian]: do you think it's time to give acx-mac80211 another try on the W701V? Aug 16 16:13:04 <[florian]> loswillios: certainly, over the past few weeks oliver has been able to test acx-mac80211 on an ar7 device and he told me he got AP mode working up to some point Aug 16 16:20:10 back in a bit, home Aug 16 18:41:14 florian * r22674 /trunk/target/linux/generic/config-2.6.35: [kernel] add missing symbol: CONFIG_UCB1400_CORE to 2.6.35 config Aug 16 19:22:28 nbd * r22675 /trunk/target/linux/ar71xx/ (7 files in 4 dirs): Aug 16 19:22:28 ar71xx: add a new driver for the ar7240 switch using swconfig. Aug 16 19:22:28 hooks directly into the ethernet driver, as MAC resets also require switch reinitializations and the switch is part of the cpu core anyway Aug 16 19:22:28 switch only tl-wr741nd (and other devices using this board file, such as tl-wr841nd) over to using this by default, as other devices are still untested Aug 16 19:22:29 fixes #7563 Aug 16 19:24:19 nbd * r22676 /branches/backfire/target/linux/ar71xx/files/drivers/net/ag71xx/ (ag71xx_ar8216.c ag71xx_main.c): ar71xx: merge ethernet driver performance fixes from trunk Aug 16 19:24:24 nbd * r22677 /branches/backfire/target/linux/ (23 files in 6 dirs): merge swconfig (+drivers) improvements from trunk Aug 16 19:24:30 nbd * r22678 /branches/backfire/target/linux/ar71xx/ (7 files in 4 dirs): ar71xx: merge the ar7240 switch driver from r22675 Aug 16 20:08:34 florian * r22679 /branches/backfire/target/linux/generic-2.6/ (4 files in 2 dirs): backport r22587 to backfire Aug 16 20:14:34 florian * r22680 /branches/backfire/target/linux/generic-2.6/patches-2.6.30/ (2 files): sync backfire with trunk for samsung flash patches, thanks maddes (#7348) Aug 16 20:27:26 florian * r22681 /branches/backfire/target/linux/ifxmips/patches-2.6.30/310-cfi0002-swap.patch: refresh patch for ifxmips Aug 16 20:54:09 florian * r22682 /branches/backfire/target/linux/generic-2.6/patches-2.6.32/ (2 files): r22680 also applies to 2.6.32 Aug 16 20:55:58 florian * r22683 /trunk/target/linux/ifxmips/patches-2.6.30/310-cfi0002-swap.patch: [ifxmips] refresh 310-cfi0002-swap.patch Aug 16 20:58:00 florian * r22684 /branches/backfire/target/linux/orion/patches/050-wrt350nv2_cfi_workaround.patch: delete now obsolete 050-wrt350nv2_cfi_workaround.patch Aug 16 21:01:20 [florian], ping Aug 16 21:02:03 <[florian]> Wipster: pong Aug 16 21:02:04 florian * r22685 /trunk/target/linux/generic/config-2.6.35: [kernel] add missing config options from #7733, thanks maddes Aug 16 21:02:23 <[florian]> Wipster: oh, forgot the patch :/ Aug 16 21:04:06 [florian], no worries no rush :) does the Fixed MDIO Bus block the actual bus like the fixed phy does? I see that Fixed MDIO gets probed before cpmac-mii Aug 16 21:05:07 <[florian]> Wipster: I do not know about fixed mdio bus Aug 16 21:07:54 [florian], ok thanks :) Aug 16 22:03:41 brb experimenting Aug 16 22:54:48 hi! Aug 17 02:54:52 is there a way to prevent a package's init.d script from being enabled by default during the build? **** ENDING LOGGING AT Tue Aug 17 02:59:57 2010