**** BEGIN LOGGING AT Mon Jul 12 02:59:57 2010 Jul 12 08:12:30 hello Jul 12 11:16:29 build #56 of ubicom32 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/56 Jul 12 11:22:12 nico * r22141 /packages/admin/syslog-ng3/Makefile: [packages] syslog-ng3: set IPv6 support explicitly, rebuild when CONFIG_IPV6 is changed, bump release number Jul 12 13:20:02 claudio * r22142 /trunk/target/linux/generic/files/drivers/pwm/ (Makefile gpio-pwm.c gpio.c): [kernel] pwm: use coherent naming to avoid conflicts Jul 12 13:20:03 claudio * r22143 /trunk/package/kernel/modules/other.mk: [kernel] Add package for gpio-pwm Jul 12 13:20:08 claudio * r22144 /trunk/package/kernel/modules/other.mk: [kernel] Add package for gpio rotary encoder Jul 12 13:42:29 ping nbd Jul 12 14:06:18 claudio * r22145 /trunk/ (63 files in 45 dirs): massive: replace occurences of .$(LINUX_KMOD_SUFFIX) with .ko after r21950 Jul 12 14:13:03 lorenzo__: pong Jul 12 14:13:12 hi Jul 12 14:13:15 hi Jul 12 14:17:42 unfortunately after some days of test the patch r22088 (511-ath9k_completion_buffer_leak.patch) does not fix my issue Jul 12 14:18:02 but I have a question about this Jul 12 14:18:39 if ieee80211_find_sta_by_hw fails Jul 12 14:20:37 Do you have to free aldo frames in tx.pending_frames ? Jul 12 14:20:45 *also Jul 12 14:20:53 erm what revision was backfire release? Jul 12 14:21:52 lorenzo__: good point Jul 12 14:23:26 how can I find the revision associated with a tag without checking it out? Jul 12 14:23:42 svn info , look for last changed rev Jul 12 14:24:02 With the patch r22088, after the problem occurred, if I bring down and up the interface the systems restarts Jul 12 14:24:26 with a kernel oops? Jul 12 14:24:35 but the pending frames are never freed Jul 12 14:25:06 no just with ifconfig wlan0 down and up Jul 12 14:25:29 no oops Jul 12 14:26:06 ok Jul 12 15:38:27 ping nbd, xMff Jul 12 15:39:52 pong cshore_ Jul 12 15:41:03 xMff: it looks like I don't have commit access to backfire....could I get you to merge a few commits? Jul 12 15:41:28 make a list of revs Jul 12 15:42:42 xMff: but just to be sure I'm doing it right, if I had access I would checkout Jul 12 15:42:42 svn+ssh://openwrt@svn.openwrt.org/openwrt/branches/backfire Jul 12 15:42:42 svn blah merge Jul 12 15:42:42 svn commit Jul 12 15:44:08 ? Jul 12 15:45:11 trunk co and branch must be within the same workcopy Jul 12 15:45:33 <{Nico}> or you have to use urls Jul 12 15:45:39 or that Jul 12 15:45:53 I prefer wc for tab-complete etc. Jul 12 15:46:20 xMff: ok, so checkout a level higher than usual? Jul 12 15:47:16 "my way": svn co -N svn+ssh://svn.openwrt.org/openwrt; cd openwrt; svn -N up branches; cd branches; svn up backfire; cd ../..; svn up trunk Jul 12 15:47:26 <{Nico}> cshore_: you can checkout openwrt using the non-recursive flag (-N) Jul 12 15:47:33 <{Nico}> xMff: damn fast Jul 12 15:47:42 <{Nico}> :) Jul 12 15:47:44 with one hand! ;) Jul 12 15:48:06 <{Nico}> xMff: impressive Jul 12 15:58:33 hehe Jul 12 16:01:18 nico * r22146 /trunk/include/kernel.mk: partially revert [22145], keep LINUX_KMOD_SUFFIX, needed to build external modules in ./packages Jul 12 16:11:49 ok, I really don't have commit access to backfire. The revisions that need merged are: Jul 12 16:11:49 r21632-r21634 and r21636-r21637 (arch-specific preinit/failsafe bug fixes) Jul 12 16:11:49 and r21987 (fixed deadlocks due to buggy use of locks in /lib/functions/block.sh) Jul 12 16:11:50 and I wonder about r22002, which fixed bugs preventing e2fsck from being executed on external root filesystem before it was mounted (when enabled) Jul 12 16:12:14 cshore_: ah btw Jul 12 16:12:29 there's a ticket suggesting swapon begfore fsck Jul 12 16:12:35 xMff: yes? Jul 12 16:12:37 could you look into it? Jul 12 16:12:45 xMff: what ticket? Jul 12 16:13:02 #7599 Jul 12 16:13:16 don't you get mails? Jul 12 16:13:26 ah right, it changed recently Jul 12 16:13:31 ping Bartman007 Jul 12 16:14:11 xMff: I hadn't check my mail yet today Jul 12 16:23:09 xMff: there not reason not to do swap first and it's just a matter of swapping two lines. In fact for block-hotplug swap is mounted before fsck already Jul 12 16:23:43 can you commit a patch and close the ticket? Jul 12 16:25:11 cshore * r22147 /trunk/package/block-mount/files/fstab.init: [package]: block-mount: Enable swap before doing fsck so that large filesystem checks have swap as well as memory (as they take large memory for large partitions). Closes #7599. Jul 12 16:25:24 ty Jul 12 16:25:56 np Jul 12 16:26:22 hmm Jul 12 16:26:36 ? Jul 12 16:26:39 actually it could mean another regression Jul 12 16:26:50 what's that? Jul 12 16:26:53 if swapfiles are on a filesystem Jul 12 16:28:10 another detail Jul 12 16:29:01 can you protect line 44-46 with [ -e /etc/fstab ] || { ... } ? Jul 12 16:30:11 xMff: problem is if hotplug is happening at the same time Jul 12 16:31:22 and it should be /tmp/fstab not /etc/fstab (just noticed) Jul 12 16:31:43 ah thats probably the root cause Jul 12 16:32:01 someone reported that symlink creation fails bc of that Jul 12 16:32:07 but probably should have that check too Jul 12 16:35:41 cshore * r22148 /trunk/package/block-mount/files/fstab.init: [package]: block-mount: Fixed two bugs in fstab.init. /etc/fstab was used where /tmp/fstab should have been, and locking was insufficiently careful and was such that it could result in deadlock when hotplug was in use. Jul 12 16:37:19 now for the swap on filesystem, we could add a new section to fstab for swap files and treat them differently than swap devices Jul 12 16:39:37 maybe its enough to just run a second swapon cycle Jul 12 16:39:40 is there an easy way to test if a device is a block device file rather than file in shell? Jul 12 16:40:11 xMff: hmmm....let me check Jul 12 16:40:35 nico * r22149 /packages/utils/ntfsprogs/Makefile: [packages] ntfsprogs: disable crypto support explicitly, use libtool fixup & PKG_INSTALL, use $(CP) instead in $(INSTALL_BIN) (perms already set by upstream 'make install'), cleanup Jul 12 16:42:44 xMff: you're right, it should work Jul 12 16:43:32 if not, test -b should test for block devices :) Jul 12 16:44:06 stintel: thanks Jul 12 17:15:41 swap on filesystem (at least on overlay) doesn't seem to be supported (segfaults) Jul 12 17:16:20 by hand, not fstab, that is Jul 12 17:17:46 hmm ok Jul 12 17:19:23 or, hmm....does swapfile swap show up on /proc/swaps? I don't see anything in there when I do mkswap /swapfile.swp (which was created with dd beforehand) Jul 12 17:19:35 yes Jul 12 17:19:53 $ cat /proc/swaps Jul 12 17:19:53 Filename Type Size Used Priority Jul 12 17:19:53 /var/swapspace file 2097144 27120 -1 Jul 12 17:20:01 hmmm Jul 12 17:20:40 don't invest too much energy into this Jul 12 17:20:54 its probably a mini_fo incompatibility Jul 12 17:21:17 I think it only works on non-ovwerlaid mounts Jul 12 17:23:25 xMff: confirmed Jul 12 17:23:44 xMff: I had another parition available, and the swapfile works on it Jul 12 17:28:52 cshore * r22150 /trunk/package/block-mount/files/fstab.init: [package] block-mount: Attempt swapon a after mounting as well as before. This ensures that swap on a filesystem is enabled. Jul 12 17:29:13 I tested it without mini_fo and it works Jul 12 17:30:27 lars * r22151 /packages/lang/python-json/ (. Makefile): [packages] Add python-json package Jul 12 19:16:25 nbd * r22152 /trunk/package/mac80211/patches/ (525-ath9k_handle_keymiss.patch 540-ath9k_aggr_fix.patch): mac80211: improve reliability of the keymiss handling fix, add another aggregation leak fix Jul 12 19:29:48 nbd * r22153 /branches/backfire/package/mac80211/patches/ (525-ath9k_handle_keymiss.patch 530-ath9k_aggr_fix.patch): [backfire] merge ath9k fixes from r22152 Jul 12 19:48:54 lars * r22154 /packages/libs/flac/ (Makefile patches/001-no_encoder.patch): [packages] libflac: Upgrade to 1.2.1 Jul 12 19:49:26 lars * r22155 /packages/libs/flac/patches/001-no-docs-and-examples.patch: [packages] flac: Disable building of docs and examples Jul 12 19:50:13 lars * r22156 /packages/multimedia/gst-plugins-good/Makefile: [packages] gst-pluings-good: Enable flac plugin Jul 12 21:11:55 hi, i'm having problems flashing au1000-au1500 to my internetbox using kernel 2.6.34.1. no problems if i use 2.6.30 :( any ideas, success stories? Jul 12 21:39:22 <[florian]> booo: 2.6.34.1 works fine on a mtx-1 here Jul 12 21:39:26 <[florian]> booo: can you be more specific? Jul 12 21:45:21 [florian], no sorry :) no serial Jul 12 21:46:50 after flashing i type go or powercycle and for a very long time nothing happen. i will try to get serial output Jul 12 21:48:14 jow: you added a hotplug script for the 3g connections Jul 12 21:48:45 it always looked to me very fragile Jul 12 21:48:54 does it work on boot? always? Jul 12 21:48:55 but thanks for your work on broken pci [florian] Jul 12 21:49:15 nunojpg: which one you mean? Jul 12 21:51:11 /etc/hotplug.d/usb/30-3g Jul 12 21:51:33 whats fragile? Jul 12 21:51:43 DOES IT WORK? Jul 12 21:51:50 for me yes Jul 12 21:51:54 I tested that a few months ago and it didn't Jul 12 21:52:02 sometimes hotplug events weren't fired on boot Jul 12 21:52:12 it didn't exist a few months ago Jul 12 21:52:25 that script... Jul 12 21:52:39 and defunct hotplug is not within the scope of that script Jul 12 21:53:18 sorry? Jul 12 21:53:48 I mean, if hotplug is broken on boot you have bigger issues than auto dialling 3g Jul 12 21:54:31 yes Jul 12 21:54:33 unless its just usb hotplug but even then its still an os bug Jul 12 21:55:43 probably things are working better now so... Jul 12 21:55:45 thanks Jul 12 21:57:41 np Jul 12 21:58:12 owh, and what does mean defunct? Jul 12 21:58:16 is it being replaced? Jul 12 21:58:30 defunctional ... broken :) Jul 12 22:22:40 log is mad Jul 12 22:22:41 http://pastebin.com/ckJYKE9L Jul 12 22:22:51 messages are mixed Jul 12 22:23:18 maybe some sync problem on logger? Jul 12 22:24:06 no Jul 12 22:24:36 this kind of output looks like pppd is communicating with a wrong tty Jul 12 22:24:51 did it choose ttyS0 by accident? Jul 12 22:25:18 who's accident? Jul 12 22:25:26 :) Jul 12 22:25:43 either 30-3g or yours Jul 12 22:26:10 that's easy to find, 5m Jul 12 22:26:53 well it's "impossible" Jul 12 22:27:20 30-3g only calls "ifup $cfg" Jul 12 22:27:39 and my only device at /etc/config/network is /dev/ttyUSB0 Jul 12 22:27:42 ok Jul 12 22:27:48 still Jul 12 22:28:32 I get output like "~^?^C^@!E^@^@^Ea^" if pppd runs on the current terminal Jul 12 22:28:49 with stuff connected to stdout Jul 12 22:29:01 therfore i suspected a script failure Jul 12 22:30:27 linux question: is possible to write to a "old" /dev/ttyUSB0 that could no longer exist? Jul 12 22:31:11 no Jul 12 22:32:23 and it is speaking to /dev/ttyUSB0, sometimes it connects Jul 12 22:32:34 so it must be a problem elsewhere Jul 12 22:33:12 does it work reliably if you unplug and replug the dongle? Jul 12 22:33:40 sometimes Jul 13 01:48:11 cshore * r22157 /packages/net/freeswitch/ (15 files in 5 dirs): [packages]: net/freeswitch: Added UCI config of extensions and incoming DIDs (to extensions only) Jul 13 01:48:51 cshore * r22158 /packages/net/freeswitch/files/uci/extensions.sh: [packages]: net/freeswitch: Fixed sed expression for escaping replaced string in the 'real' sed command. **** BEGIN LOGGING AT Tue Jul 13 15:10:14 2010 Jul 13 15:51:05 ping cshore Jul 13 16:00:44 pong acoul Jul 13 16:03:19 hey Daniel, check this: http://openwrt.pastebin.com/AbMdX3iE Jul 13 16:03:34 line 119 Jul 13 16:03:57 https://dev.openwrt.org/browser/trunk/target/linux/adm5120/base-files/lib/preinit/05_preinit_do_adm5120.sh?rev=22167 Jul 13 16:06:26 hmmm...why is proc not mounted? Jul 13 16:06:55 proc is not mounted before preinit Jul 13 16:06:55 wait: I see. Jul 13 16:07:31 the error message is from the loading of diag.sh in /etc/preinit, not from the file you showed Jul 13 16:07:43 hmmm Jul 13 16:08:43 it *should* be harmless, but the only way to get rid of is to not load the lib/adm5120.sh in diag.sh Jul 13 16:08:52 or move when diag.sh is loaded Jul 13 16:10:33 I am having some issues with jffs2/lzma and I was wondering if this may be related: http://openwrt.pastebin.com/M401m1Yy Jul 13 16:11:56 acoul: let me check the source code Jul 13 16:12:06 good Jul 13 16:15:27 acoul: is this in a clean buildroot? Jul 13 16:15:44 yes Jul 13 16:16:09 are you using udev or hotplug? Jul 13 16:16:52 what kernel version? Jul 13 16:17:25 ok, I see, .34 Jul 13 16:17:34 http://openwrt.pastebin.com/JA6xRyWm Jul 13 16:18:06 do you know if it's using devfs? Jul 13 16:18:20 squashfs is ok Jul 13 16:18:24 so let me check Jul 13 16:20:17 no devfs Jul 13 16:20:30 jow * r22168 /trunk/target/linux/generic/ (6 files in 6 dirs): [kernel] export SSB fallback SPROM - required for wl.ko Jul 13 16:22:37 hmmm....didn't the .33 or .34 enable devfs2 by default in the kernel or somesuch Jul 13 16:23:33 because the path /dev/ptmx should normally exist Jul 13 16:24:58 hotplug2 should create it in /lib/preinit/30_device_fs_daemons Jul 13 16:26:30 <[florian]> the kernel can maintain a /dev/ using devtmpfs Jul 13 16:26:30 I have access to the /proc/config.gz on the squashfs image Jul 13 16:26:51 oh, devtmpfs, that's what I was thinking of Jul 13 16:27:14 CONFIG_DEVTMPFS is not set Jul 13 16:27:40 the hotplug isn't creating the devices Jul 13 16:27:43 *then Jul 13 16:28:23 or the system is getting confused and not using hotplug Jul 13 16:28:32 <[florian]> or the rootfs is not correctly decompressed Jul 13 16:28:43 <[florian]> because acoul is trying to use a lzma compressed jffs2 rootfs Jul 13 16:29:16 ah, yeah, there's not reason in the preinit code it shouldn't work Jul 13 16:29:32 BTW, I have an rb133 working just fine with jffs2/lzma Jul 13 16:29:35 Linux version 2.6.33.5 (alex@motard) (gcc version 4.4.4 (GCC) ) #1 Tue Jun 22 16:15:52 EEST 2010 Jul 13 16:29:52 <[florian]> acoul: ok so the build system creates working images then Jul 13 16:30:26 [florian]: btw, hello. It's been a while. Jul 13 16:30:44 <[florian]> cshore: yeah, I have been fairly busy lately, it's not over actually :) Jul 13 16:30:50 this is a compex WP54G, an adm5120 too Jul 13 16:32:20 Daniel here is the squashfs bootlog: http://openwrt.pastebin.com/AbMdX3iE Jul 13 16:34:19 acoul: definitely a lzma/jffs2 issue then Jul 13 16:35:53 maybe an arch-specific issue (endianness perhaps?) Jul 13 16:36:15 <[florian]> cshore: rb133 and wp54 are the same hardware actually Jul 13 16:36:17 <[florian]> that's strange Jul 13 16:37:29 Daniel this is the lzma/jffs2 build stage: http://openwrt.pastebin.com/LsaNr620 Jul 13 16:37:30 do they have different consoles maybe? Jul 13 16:37:46 <[florian]> both are on ttyS0 Jul 13 16:38:17 besides the problems he's having is with the pseudo-terminal Jul 13 16:38:24 which is weird Jul 13 16:39:47 acoul: same tree used for the rb? Jul 13 16:40:01 yes, but a few weeks older Jul 13 16:40:17 but no proprietary code Jul 13 16:41:05 rb133 lzma/jffs2 is on r21871 Jul 13 16:44:42 what kernel on the rb? Jul 13 16:45:58 Linux version 2.6.33.5 (alex@motard) (gcc version 4.4.4 (GCC) ) #1 Tue Jun 22 16:15:52 EEST 2010 Jul 13 16:46:32 acoul: so probably something to do with the .34 kernel Jul 13 16:48:42 acoul: does it work with .34 kernel without lzma? Jul 13 16:48:42 before pushing patches on the tree I always test them on actual hw Jul 13 16:49:01 rb133 has also booted fine on 2.6.34 on lzma/jffs2 Jul 13 16:49:36 I just pulled out of the draw this dusty WP54G and bumped onto this issue Jul 13 16:50:01 does this wp54g work without lzma? Jul 13 16:50:20 this is my next test Jul 13 16:50:31 I was hoping to avoid it though :) Jul 13 16:52:02 juhosg * r22169 /trunk/tools/firmware-utils/src/mktplinkfw.c: Jul 13 16:52:02 tools/firmware-utils: enhance mktplinkfw utility Jul 13 16:52:02 Attached patch enhances the mktplinkfw utility: Jul 13 16:52:02 1. existing firmware files in TP-Link format can now be inspected, showing eg. the used offsets Jul 13 16:52:02 2. optionally the kernel and rootfs parts can be extracted (saves calculations for dd) Jul 13 16:52:03 3. when generating firmwares, kernel entry point, load address and rootfs offset can be overwritten (for experienced users) Jul 13 16:52:04 Patch-by: Pieter "Fate" Hollants Jul 13 16:52:04 juhosg * r22170 /trunk/target/linux/ramips/ (rt288x/config-2.6.34 rt305x/config-2.6.34): ramips: sync kernel config Jul 13 16:52:06 juhosg * r22171 /trunk/target/linux/ramips/files/arch/mips/ralink/common/gpio.c: ramips: protect GPIO register access with a spinlock Jul 13 16:52:11 juhosg * r22172 /trunk/target/linux/ramips/files/drivers/net/ramips.c: (log message trimmed) Jul 13 16:52:11 ramips: Fix bridging in ramips ethernet driver Jul 13 16:52:11 Bridging between the ramips ethernet driver and rt2800pci was somewhat Jul 13 16:52:11 broken. Frames received by the ethernet driver which were passed to the Jul 13 16:52:11 wifi driver for transmission were sometimes corrupted or sent out with Jul 13 16:52:11 huge delays. Jul 13 16:52:12 The reason for this is the missing assignment of skb->tail in the ramips Jul 13 16:52:19 a 2.6.34.1 issue possibly? I will test this on an rb133 later Jul 13 16:55:56 anyway, need to hit the road Jul 13 16:55:59 jow * r22173 /trunk/package/broadcom-wl/ (3 files in 2 dirs): Jul 13 16:55:59 [package] broadcom-wl: Jul 13 16:55:59 - update to v5.10.56.27.3 Jul 13 16:55:59 - get fallback SPROM from SSB if neither on-device SPROM nor nvram is available Jul 13 16:55:59 - working brcm63xx support Jul 13 16:56:06 talk to you soon Jul 13 17:03:58 jow * r22174 /trunk/target/linux/brcm63xx/ (2 files in 2 dirs): [brcm63xx] add required exports for wl.ko Jul 13 17:21:45 jow * r22175 /trunk/package/broadcom-wl/Makefile: [package] broadcom-wl: fix md5sums Jul 13 18:45:31 nbd * r22176 /trunk/Config.in: enable ipv6 support for packages by default Jul 13 18:46:27 nbd * r22177 /branches/backfire/Config.in: [backfire] enable ipv6 support for packages by default (r22176) Jul 13 19:22:49 Just FYI (nbd, jugosh, et al): latest backfire is finally working very reliable in ath9k (WRT160NL) Jul 13 19:23:47 nice Jul 13 20:17:04 great times ahead :) Jul 13 21:06:55 time to reset my router's uptime to check the ath9k improvements it seems :) Jul 13 21:07:04 23:06:30 up 43 days, 23:37, load average: 0.00, 0.00, 0.00 **** ENDING LOGGING AT Wed Jul 14 02:59:56 2010