**** BEGIN LOGGING AT Fri May 28 03:00:14 2010 May 28 03:26:06 build #54 of etrax is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/54 May 28 06:06:40 hello May 28 10:25:50 build #54 of ifxmips is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/ifxmips/builds/54 May 28 10:46:14 build #75 of ppc40x is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/ppc40x/builds/75 May 28 10:52:06 jow * r21597 /trunk/package/base-files/files/lib/network/config.sh: [package] base-files: revert r21595 May 28 10:53:11 jow * r21598 /branches/backfire/package/base-files/files/lib/network/config.sh: [backfire] revert r21596 May 28 12:00:25 nico * r21599 /trunk/toolchain/gcc/ (15 files in 2 dirs): toolchain/gcc: add support for gcc-4.3.5 May 28 12:00:33 nbd: ping May 28 12:01:26 rtz2: pong May 28 12:02:37 nbd: did you have time to look at the mmap problem? May 28 12:02:43 nope May 28 12:04:07 http://downloads.openwrt.org/sources/openssh-5.5p1.tar.gz doesn't exist May 28 12:04:32 someone please update it, I have ftp blocked and this is the only mirror the buildroot can access... May 28 12:05:14 nbd: in this case, do you really think, that changing the len value in in the trx header to a value smaller then the kernel length is likely to create a problem in the future? May 28 12:06:33 nbd: because the first version of the mtd patch works, and it would be rather nice, if this could get comitted some time in the not-too-distant future May 28 12:08:56 rtz2: yes, i think this could cause some issues May 28 12:11:56 nbd: ok, then if I would manage to track down the mmap bug, could you review the rest of the mtd patch? May 28 12:12:06 will do May 28 12:12:52 nbd: ok May 28 12:20:27 <{Nico}> nunojpg: done > http://downloads.openwrt.org/sources/openssh-5.5p1.tar.gz May 28 12:33:15 build #73 of ppc44x is complete: Failure [failed shell_6] Build details are at http://tksite.gotdns.org:8010/builders/ppc44x/builds/73 May 28 13:41:16 heh, now even wifi works May 28 13:50:49 nico * r21600 /trunk/toolchain/gcc/ (13 files in 2 dirs): toolchain/gcc: add support for gcc-4.4.4 May 28 15:12:36 Hello. May 28 15:13:30 I have a device which run a bootloader named i3Boot. Anyone know anything about this? May 28 15:15:03 The only commands I can get it to accept is 'reboot' and 'i3boot' (which just loads the default kernel). May 28 15:15:28 Hi guys, I am thinking about contributing specific ipk packages for various 3G USB dongles. The crucial part is writing correct usb hotplug scripts. My question is whether everything gets executed within the /etc/hotplug.d/usb/ directory in lexicographic order or only 10-usb upon plugging something in? May 28 15:15:43 epointsystem: everything May 28 15:16:13 Thanks May 28 15:33:45 nico * r21601 /trunk/toolchain/eglibc/Config.in: toolchain/eglibc: allow building eglibc-2.11 with gcc-4.5.x May 28 16:49:27 kerneis * r21602 /packages/ipv6/ahcpd/files/ahcpd.config: May 28 16:49:27 [package] ahcpd: fix typo in config file May 28 16:49:27 Signed-off-by: Gabriel Kerneis May 28 16:54:59 epointsystem: regarding your plans for an ipk for 3g dongles, do you mean integrating usb_modeswitch with that? May 28 16:57:32 nbd: you still here? May 28 17:02:11 xMff: exactly May 28 17:02:28 epointsystem: great, it is long overdue imo :) May 28 17:02:39 Writing proper configuration for usb_modeswitch, setting the right interface, etc. May 28 17:03:14 you could add a .conf for each quirk in /etc/usb_modeswitch/xyz.conf and then select them based on vendor and product id May 28 17:03:26 Yes, that's what I am doing. May 28 17:03:58 Except that it's /etc/usb-modeswitch-xyz.conf May 28 17:04:16 I think you should consider a subdir, its less clutter in /etc then May 28 17:04:25 can be easily dozens of files I think May 28 17:04:25 Sure. May 28 17:04:53 While we are at it, how should I go about /etc/uci-defaults ? May 28 17:05:03 what do you mean specifically? May 28 17:06:37 The way I would like to do it is that I add an additional interface called mobile (set network.mobile=interface), configure it, and add it to the firewall zone wan. May 28 17:06:44 (if there is such a zone, of course) May 28 17:07:00 That's a script that belongs to uci-defaults, IMHO. May 28 17:07:02 hmm May 28 17:07:24 I think usb_modeswitch should not setup a default iface config, that is comgt's job May 28 17:08:24 Neither should. I don't want to touch either package. I would like to make a bunch of packages that depend on both usb-modeswitch and comgt. May 28 17:08:35 ah ok May 28 17:08:44 uci-defautls is the right script for this May 28 17:09:00 /etc/uci-defaults/setup-3g.sh May 28 17:09:09 then do a uci batch operation in there May 28 17:09:41 But do I simply add these to the files section of the package or do I do the postinst magic that some packages do? May 28 17:10:14 it should be enough to just put it in files May 28 17:10:40 if the package is builtin the script is executed with firstboot May 28 17:10:51 if it is opkg installed it will run after the next reboot May 28 17:11:34 And removed, right? May 28 17:11:38 yes May 28 17:12:48 if you want to apply the uci-defaults scipts immediately on opkg install (without needing a reboot) use a script like this: http://luci.subsignal.org/trac/browser/luci/trunk/applications/luci-p2pblock/ipkg/postinst May 28 17:12:53 as postinst May 28 17:12:56 So, here's the plan: each of these packages will depend on: usb-modeswitch (if necessary, but usually it is), comgt and the appropriate kernel module (e.g. kmod-usb-serial-option). May 28 17:13:19 the [ -n "${IPKG_INSTROOT}" ] check ensures that the postinst only runs for opkg install and not when executed by buildroot May 28 17:13:23 Well, that's what I was asking. Should I just put in the file, or do that thing? May 28 17:13:30 both May 28 17:13:48 So, you think that the package should work without needing a reboot, right? May 28 17:14:04 if you're fine with prompting users to reboot after installing the package with opkg, you can leave out the postinst May 28 17:14:14 its basically only needed if you want to apply uci-defaults immediately May 28 17:14:34 well yes I think it should May 28 17:14:45 Then it will. :-) May 28 17:14:47 epointsystem: you want to have a much improved dialup script for comgt? May 28 17:14:58 because a user might start to configs his stuff manually, then reboot and then he'll wonder why stuff was added to the config May 28 17:16:16 Good point. May 28 17:17:17 rtz2: not really. The one in current OpenWrt works fine with all the hardware I threw at it (including phones). May 28 17:21:02 epointsystem: well, I had some problems with it, but ok May 28 17:43:35 Care to review the sources? https://www.epointsystem.org/trac/vending_machine/browser/hotspot/trunk/huawei-e175x May 28 17:43:41 Alternatively: svn co https://www.epointsystem.org/svn/vending_machine/hotspot/trunk/huawei-e175x May 28 17:50:41 epointsystem: what about placing the files in for example /etc/usb-modeswitch/12d1/1446/0.conf ? May 28 17:51:30 epointsystem: then you can do [ -f /etc/usb-modeswitch/$PRODUCT.conf ] && usb_modeswitch -c /etc/usb-modeswitch/$PRODUCT.conf May 28 17:52:26 So, are you suggesging to make a general package upon which all the particular ones depend? May 28 17:52:53 yes May 28 17:53:16 a kind of quirk database for usb_modeswitch May 28 17:53:39 I'll think about it, but keep in mind that not all modems need usb_modeswitch. May 28 17:53:44 yes May 28 17:53:52 the proposed solution above would handle that nicely May 28 17:54:11 if no file exists for that particular vendor/id combo, modeswitch is not launched May 28 17:54:53 and it allows you to simply define new entries by placing files without having to extend if/else blocks in some script May 28 17:55:41 as for the ifup May 28 17:55:49 this can be solved in a generic way too May 28 17:56:02 can you provide me with a "env" dump when a dongle is plugged? May 28 17:56:40 Okay, now I'm off to meet Mr. Kaloz and buy him a drink. Thanks for the suggestions! I'd appreciate emails to nagydani@epointsystem.org with all requests. May 28 17:57:12 okay. have fun May 28 19:45:12 juhosg * r21603 /trunk/target/linux/ramips/ (rt288x/config-2.6.32 rt305x/config-2.6.32): ramips: sync 2.6.32 config May 28 19:45:18 juhosg * r21604 /trunk/target/linux/ramips/patches-2.6.34/ (4 files): ramips: refresh 2.6.34 patches May 28 19:45:22 juhosg * r21605 /trunk/target/linux/ramips/ (rt288x/config-2.6.34 rt305x/config-2.6.34): ramips: add missing 2.6.34 config files May 28 19:45:27 juhosg * r21606 /trunk/target/linux/ramips/ (2 files in 2 dirs): ramips: use COMMAND_LINE_SIZE directly May 28 19:45:30 juhosg * r21607 /trunk/target/linux/ramips/ (3 files in 2 dirs): ramips: merge slab patch May 28 19:48:07 nbd * r21608 /trunk/package/grub/ (4 files in 2 dirs): grub: compile fixes for Mac OS X (based on patches by Jukka Ylitalo) May 28 19:48:23 nbd * r21609 /trunk/target/linux/x86/image/Makefile: x86: build fully functional images on Mac OS X, now that grub is fixed May 28 19:48:27 nico * r21610 /trunk/toolchain/eglibc/ (Config.in Config.version): toolchain/eglibc: update to latest revision for 2.9, 2.10 & 2.11 May 28 20:06:47 when there are changes in toolchain, does it need to be recompiled? and if is there a easyes way then make clean && make distclean, as this cleans packages feeds May 28 20:07:27 most of the time it does not need to be recompiled as long as the old one still works May 28 20:07:35 if you want to clean it, use make dirclean world May 28 20:07:41 this will rebuild everything May 28 20:16:12 ok thnx :) May 28 20:37:58 nico * r21611 /trunk/toolchain/eglibc/ (Config.in Config.version Makefile config/Config.in): toolchain/eglibc: add preliminary support for eglibc-2.12 May 28 20:39:25 build #53 of goldfish is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/goldfish/builds/53 May 28 20:47:56 nbd: still here? May 28 20:48:02 yes May 28 20:49:48 nbd: about mtd, how useful are the -e (erase) option and the ability to feed an image via stdin to mtd? May 28 20:50:04 stdin is important May 28 20:50:14 nbd: because removing one or both of them would make some thing easier May 28 20:50:15 it is used by sysupgrade on atheros and ar71xx May 28 20:50:26 removing -e would be acceptable May 28 20:50:40 but stdin is quite important, yes May 28 20:50:46 ok May 28 20:51:27 do you see a problem with writing the image via a mmaped mtdblock device instead of a raw mtd device? May 28 20:51:31 wouldn't flashing a rootfs without -e leave stray jffs2 markers around and stuff like that? May 28 20:51:39 xMff: nope May 28 20:52:18 xMff: the image written contains a jffs2 start maker and if there is some stuff later, it shouldn't matter May 28 20:52:39 xMff: but I will run a few tests to make sure May 28 20:52:54 so the search for jffs2 is from begin to end sequentially? May 28 20:52:59 *for jffs2 marker May 28 20:53:09 and first one wins? May 28 20:54:58 yes May 28 20:55:35 k May 28 20:59:35 aaaannnd I lost my inet connection again May 28 20:59:40 xMff: you mean the deadcode marker or the 0x1986 marker from the actual jffs2 fs? May 28 21:00:38 anyone patched up gcc 4.5.0 to workaround http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43987 causing sed to segfault? May 28 21:01:15 swalker: I tried, it's not straight forward May 28 21:01:47 swalker: there were I number of other changes in the files affected by the patch May 28 21:02:03 swalker: you could try a gcc svn checkout May 28 21:02:13 but I guess it's easier to wait for 4.5.1 May 28 21:02:22 yeah, I tried May 28 21:03:53 swalker: I don't think it's worth the effort to backport it May 28 21:08:23 rtz2: I meant the deadcode marker May 28 21:08:48 rtz2: since all flash beginning from deadcode is erased, stray jffs2 leftovers should be no issue May 28 21:10:58 xMff: it should, let me check May 28 21:11:06 it does May 28 21:11:12 I am like 99% sure May 28 21:14:06 xMff: the kernel simply ignore the marker May 28 21:14:39 xMff: the split happens on the first eraseblock after the end of the squashfs May 28 21:15:30 xMff: deadcode is only important for userspace, where it has to be at the start of the partition anyway May 28 22:03:33 jow * r21612 /trunk/package/6in4/ (Makefile files/6in4.hotplug files/6in4.sh): May 28 22:03:33 [package] 6in4: May 28 22:03:33 - support to automatically determine the local endpoint address from the current IPv4 default gateway May 28 22:03:33 - support updating the tunnel endpoint for he.net May 28 23:01:37 does anybody know how to but a wav-281 into serial upload mode? May 29 00:05:15 lars * r21613 /trunk/target/linux/xburst/patches-2.6.34/052-rtc.patch: [xburst] rtc: Put rtc Kconfig entry into the correct section May 29 00:06:21 lars * r21614 /trunk/target/linux/xburst/patches-2.6.34/051-fb.patch: [xburst] fb: Fix pseudo palette for non 32bit color depth. May 29 00:07:02 lars * r21615 /trunk/target/linux/xburst/patches-2.6.34/060-jzcodec.patch: [xburst] jzcodec: Enable support for S8 format May 29 00:08:09 jow * r21616 /trunk/target/linux/brcm-2.4/image/Makefile: [brcm47xx] Rename .trx2 image to openwrt-wrt54g3gv2-vf-squashfs-sysupgrade.bin and openwrt-wrt54g3gv2-vf-squashfs.bin to openwrt-wrt54g3gv2-vf-squashfs-factory.bin May 29 00:09:16 lars * r21617 /trunk/target/linux/xburst/patches-2.6.34/ (001-core.patch 005-add-qi_lb60-board-support.patch): May 29 00:09:16 [xburst] core fixes May 29 00:09:16 * Mask timer irq on startup May 29 00:09:16 * Fix dma width constants May 29 00:09:16 * Do not try to ack intc irqs May 29 00:09:53 lars * r21618 /trunk/target/linux/xburst/patches-2.6.34/061-asoc.patch: [xburst] ASoC: Fix and cleaup dma config May 29 00:12:12 lars * r21619 /trunk/target/linux/xburst/patches-2.6.34/056-udc.patch: (log message trimmed) May 29 00:12:12 [xburst] udc: Fix recursive spinlocks May 29 00:12:12 Driver had recursive spinlock locking: 1. jz4740_queue() acquires lock 2. done() May 29 00:12:12 is called in chain: May 29 00:12:12 jz4740_queue()->jz4740_ep0_kick()->jz4740_ep0_in()->write_fifo_ep0()->done() and May 29 00:12:13 it tries to acquire same lock. 3. Deadlock. May 29 00:12:14 Signed-off-by: May 29 00:33:42 jow * r21620 /branches/backfire/target/linux/brcm-2.4/image/Makefile: [backfire] merge r21616 May 29 00:53:41 build #51 of iop32x is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/iop32x/builds/51 May 29 02:00:07 build #47 of ramips is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ramips/builds/47 **** ENDING LOGGING AT Sat May 29 02:59:57 2010