**** BEGIN LOGGING AT Thu May 13 02:59:56 2010 May 13 04:48:47 build #44 of s3c24xx is complete: Failure [failed compile_10] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/44 May 13 06:02:31 should any changes, no matter how small be submitted via the method described here https://dev.openwrt.org/wiki/SubmittingPatches ? May 13 06:02:50 zombik: what's the change? May 13 06:03:36 zombik: but it really does make life a lot easier May 13 06:03:45 zombik: to have a diff May 13 06:03:56 its about the size of the mtd partition for kernel in https://dev.openwrt.org/browser/trunk/target/linux/ar71xx/files/drivers/mtd/nand/rb4xx_nand.c May 13 06:04:43 zombik: then yes, the devs for ar71xx will need to review the request May 13 06:05:15 ".size = (4 * 1024 * 1024) - (256 * 1024)" is too small for backfire, ".size = (8 * 1024 * 1024) - (256 * 1024)" works tho May 13 06:06:09 cshore: ok i'll try to submit the patch correctly May 13 06:06:12 zombik: I'm afraid I don't know ar71xx to comment May 13 06:06:44 zombik: but that looks like flash size, which really is 4 on some devices May 13 06:07:34 zombik: I think the real patch is something else May 13 06:08:32 zombik: self-compiled or the downloads? May 13 06:08:54 cshore: you mean, that there should be a better way, than increasing the size of kernel partition May 13 06:09:16 zombik: I mean I'm not sure you're actually changing only the kernel partition May 13 06:09:31 cshore: self compiled, out of the https://dev.openwrt.org/browser/branches/backfire branch May 13 06:11:41 zombik: hmm....it looks like you have an 8MB flash device, and the size you're changing is the size of the flash device. Maybe needs another profile for your device so you can use the 8MB flash instead of only 4MB May 13 06:13:13 cshore: the device itself (routerboard 433AH) has 512 MiB flash. according to the datasheet, it should be 64MiB, but dmesg says clearly 512Mib May 13 06:13:37 zombik: oh1 May 13 06:13:39 oh! May 13 06:13:48 zombik: ok, I'm used to smaller devices May 13 06:14:08 zombik: then yeah, probably that is the kernel partition May 13 06:14:47 zombik: can you determine if 5 or 6 would work? May 13 06:15:21 zombik: and in your submission, say what modules, options your doing differently from the defaults May 13 06:15:48 zombik: i.e. are you customizing the kernel, or only using menuconfig? May 13 06:18:25 cshore: i enabled ccache and let it build the gdb. is it worthy of mention? May 13 06:18:37 zombik: because the concern will be for smaller devices in the same profile; yes May 13 06:19:02 cshore: ok, i'll try the smaller values May 13 06:19:06 zombik: the gdb might increase the size of the kernel May 13 06:19:24 zombik: if you're including debugging symbols May 13 06:34:16 now, that i think about it... it would also be a good idea if https://dev.openwrt.org/browser/trunk/target/linux/ar71xx/base-files/sbin/wget2nand would not fail silently if there is no space on partiton to fit the kernel... May 13 06:37:11 ok, i'll test the different sizes of the kernel partition now, be right back May 13 06:48:56 cshore: well, it seemed to fit, but it would not boot... May 13 06:49:46 cshore: i guess i'll test it a little more later, before submitting. May 13 06:50:09 zombik: ok, thanks May 13 06:50:27 cause compiling on a netbook is taking a little bit to long to be fun :) May 13 06:50:43 heh May 13 06:51:01 zombik: yeah my dual core is slower than I like May 13 06:51:19 for compiling that is May 13 06:51:51 cshore: its never fast enough :) May 13 06:52:26 :) May 13 08:08:22 good morning May 13 08:52:32 good morning May 13 11:07:55 acoul: ping May 13 11:36:41 lars * r21441 /trunk/package/compcache/ (Config.in Makefile): [package] compcache: Kernel module and userspace daemon in seperate packages. May 13 13:55:30 cshore: testing your fix for WDS in broadcom.sh in a moment May 13 13:55:48 Then I can try failsafe in brcm 2.6 after that too May 13 13:57:22 Mr_Smoke nice for 2.6 test, i was thinking this weekend to put serial on my asus wl500gpv1 before that i dont wanna play that hard :D May 13 13:57:44 Hehe :) May 13 13:57:51 What should I look out for ? May 13 13:57:59 I mean what are the main differences to expect May 13 13:58:08 between brcm-2.4 and brcm47xx ? May 13 13:59:15 (failsafe-wise obviously) May 13 14:07:35 cshore: ok this is great May 13 14:07:42 Not only did your fix work May 13 14:07:59 But also, Now my WDS on backfire/trunk can work again with another unit running 7.09 May 13 14:08:08 And that's a psk2-encrypted WDs, mind you :) May 13 14:08:44 I could buy you a beer right now, if you were in Paris :) May 13 14:13:11 crow: oh sorry, wrong HL here, i thought he had replied :) May 13 14:13:49 Which means I'm not sure we're on the same issue here : what tests did you intend to run ? Maybe I can do yours too May 13 14:20:12 can a sysupgrade image be flashed directly with mtd, or does it contain extra magic? May 13 14:21:05 you mean a .trx ? May 13 14:21:19 Or else what exactly do you call a "sysupgrade image" ? May 13 14:21:35 openwrt-ar71xx-generic-wndr3700-squashfs-sysupgrade.bin May 13 14:22:34 the bin files are for when you flash using the bootloader's TFTPd I believe May 13 14:22:52 Is your router running OpenWRT already ? May 13 14:23:12 yeah, I'm used to using mtd, just trying to learn about sysupgrade May 13 14:23:45 basically, you just grab the trx image for your target May 13 14:23:51 I usually put mine in /tmp/ May 13 14:24:03 Then just "sysupgrade /tmp/openwrt-image-blablabla.trx" May 13 14:24:05 And voila :) May 13 14:24:14 It's *that* simple :) May 13 14:25:29 hehe, simple is good, I just like to know how things work under the hood May 13 14:25:44 I'll have to open the script and spend some time with it May 13 14:27:33 thanks Mr_Smoke May 13 14:31:51 Anytime :) May 13 15:47:59 nbd: ping ? May 13 16:04:58 Actually, anyone with pretty good insight on broadcom wireless is welcome to help May 13 16:05:02 https://dev.openwrt.org/ticket/7335 May 13 16:12:05 Mr_Smoke: I have a good insight into the ssb May 13 16:12:05 Mr_Smoke: depending on your question, I may be able to help May 13 16:12:16 sorry, got disconnected May 13 16:21:22 Hm May 13 16:21:34 What was that again ? May 13 16:21:50 rtz2: ssb ? May 13 16:22:46 Mr_Smoke: it's used to connect the different components on broadcom chips May 13 16:24:03 rtz2: well, I've posted a link to a ticket I opened, dunno if you've seen it ? May 13 16:26:22 nope May 13 16:27:30 Basically, when using the default broadcom.sh, WDS will *not* work *unless* there is another AP defined in /etc/config/wireless May 13 16:28:09 With cshore's patch ( see https://dev.openwrt.org/ticket/7335 ), WDS will *not* work *unless* there is *no* other AP defined in /e/c/wireless May 13 16:28:18 doesn't look like an ssb question, sorry May 13 16:28:21 I am not sure how to make both cases work May 13 16:28:23 Ok :) May 13 17:12:23 Hi May 13 17:13:10 OHAI May 13 17:13:50 I have noticed that the new httpd (the one from luci instead of busybox) does not follow symlinks. I would like it to serve a directory from a mounted drive. How should I proceed? May 13 17:14:06 Just putting a symlink into /www doesn't cut it anymore. May 13 17:14:44 Is there some documentation about /etc/config/lucid ? May 13 17:52:17 hm, some gdb guru here? I want to watch a variable and print out the new state on each write to it. I do _not_ want to halt the program on changes. watch always seems to halt the program. Is there a way to auto-continue a program after a breakpoint occured? May 13 17:55:57 The ignore count of continue is not an option, because I need a print of the new status on every breakpoint. I just want it to automatically continue immediately May 13 17:58:54 Ok, I found it. It's the "commands" command. May 13 17:59:09 It lets you queue commands to execute automatically on a breakpoint. May 13 18:04:23 Okay, I have figured it (lucid configuration) out. it's actually very nice and even supports name-based virtual hosting. May 13 18:24:31 juhosg * r21442 /trunk/package/zlib/Makefile: May 13 18:24:31 Fix zlib alternate URL May 13 18:24:31 There is no zlib project on Sourceforge, zlib files are in libpng project May 13 18:24:31 Signed-off-by: Gilles Espinasse g.esp@free.fr May 13 19:19:34 I am trying to make sense of this documentation: http://nuwiki.openwrt.org/doc/techref/block_mount May 13 19:20:08 It is still not clear, what actually triggers the block-hotplug mechanism. May 13 19:21:12 The problem I am observing is that plugging the micro-sd card does not actually result in running /etc/hotplug.d/block/40-mount May 13 19:23:40 I can mount the device by hand (mount from command line), so all the drivers are correct. The fstab configuration also seems to be correct. But 40-mount is not even started (I added some logging for debugging purposes and those commands simly do not get executed). What am I doing wrong? May 13 19:24:33 The fstab configuration confirms both the examples in the documentation and its visualization in luci also seems correct. May 13 19:59:15 Moreover: after /etc/init.d/fstab start everything is mounted as specified. But plugging in the device does not trigger the action. May 13 20:00:24 Hmm, fstab runs before usb. That might be wrong. May 13 21:01:54 epointsystem: I think the problem is the micro-sd card does not generate a kernel hotplug event May 13 21:02:26 epointsystem: or else the type of device we're looking for in hotplug is wrong May 13 21:04:27 epointsystem: the driver probably needs to be enhanced to do hotplugging, or you have to have the device plugged in so that the fstab run mounts it May 13 21:08:55 cshore: there is one thing that contradicts you. The block device in /dev (/dev/sda) gets created automagically, which means that the kernel hotplug event does happen. May 13 21:09:16 epointsystem: for sd? May 13 21:09:32 epointsystem: is it built into the kernel or a kmod? May 13 21:09:39 kmod May 13 21:15:54 Linux version 2.6.32.9 May 13 21:16:30 epointsystem: but do insert/remove create/remove the device? May 13 21:17:35 insert does create it, remove does not remove. May 13 21:18:21 epointsystem: when it's created do you see a hotplug event? May 13 21:19:10 How do I catch it? May 13 21:20:16 Honestly, I am not quite sure how hotplug architecture works and what exactly creates the block device. May 13 21:21:07 I was under the impression that it is created by some userland process (running mknod) upon receiving a hotplug event. May 13 21:22:35 How does /dev get populated in OpenWrt with 2.6 kernels? Previously, I have only used 2.4 series (because of the broadcom driver issue) and those used udev for this purpose. But that's not how 2.6 works. May 13 21:23:36 epointsystem: that's right May 13 21:24:39 epointsystem: 2.4 uses a userland (not udev) too, unless you specifically use udev instead of hotplug (the default is hotplug, both for 2.4 and for 2.6) May 13 21:25:19 epointsystem: hotplug being in 2.4 a userland executable called by the kernel, and in 2.6 a daemon May 13 21:25:49 that is a daemon that responds to kernel hotplug events May 13 21:28:05 Correct. May 13 21:28:50 So, if the device node gets created, it is proof positive that a hotplug event occurs, right? May 13 21:29:41 Yet, /etc/hotplug.d/block/40-mount does not get called. May 13 21:31:29 epointsystem: I'm not sure. Is the device not there on boot, but created when you plug in the card? May 13 21:33:04 you might want to look at /etc/hotplug.rules (or whatever it's called; let me check) May 13 21:34:05 /etc/hotplug2-common.rules and /etc/hotplug2.rules for after boot and /etc/hotplug2-init.rules for preinit May 13 21:34:51 I think that is the case, but let me check again. May 13 21:36:36 epointsystem: in /etc/hotplug2-common.rules uncomment the uncomment me section to see the hotplug events that are being sent May 13 21:36:45 they got to the syslog May 13 21:38:04 Actually, no. I was wrong. The device does not get created, if it is not there on boot. May 13 21:38:12 ah, ok May 13 21:38:36 epointsystem: so there is no kernel hotplug event (on boot it also does coldplug) May 13 21:38:46 Wait a minute. May 13 21:39:08 On a second attempt, it did get created. May 13 21:39:16 (pulled it out and plugged back in) May 13 21:39:44 but not mounted? May 13 21:39:53 Not mounted. May 13 21:40:07 try the uncomment thing and see what it reports May 13 21:40:44 Where to uncomment? May 13 21:42:06 BTW, here's the relevant part from dmesg: May 13 21:42:07 scsi 4:0:0:0: Direct-Access HUAWEI SD Storage 2.31 PQ: 0 ANSI: 2 May 13 21:42:07 usb-storage: device scan complete May 13 21:42:07 sd 4:0:0:0: [sda] Attached SCSI removable disk May 13 21:42:07 usb-storage: device scan complete May 13 21:43:43 n /etc/hotplug2-common.rules uncomment the uncomment me section to see the hotplug events that are being sent May 13 21:45:17 There isn't a single comment. May 13 21:45:37 what version ? May 13 21:46:18 in trunk: # uncomment me to log hotplug events May 13 21:46:18 # DEVPATH is set { May 13 21:46:19 # nothrottle May 13 21:46:19 # exec logger -s -t hotplug -p daemon.info "name=%DEVICENAME%, path=%DEVPATH%" May 13 21:46:19 # } May 13 21:46:37 hotplug2 - 1.0-beta-1 May 13 21:47:16 epointsystem: but trunk, backfire, kamikaze? May 13 21:47:29 backfire May 13 21:47:48 epointsystem: ah, it must have been added later then May 13 21:51:08 Wow, hotplug2 has been under fairly intense tweaking in the past 7 weeks according to SVN logs... May 13 22:02:31 I added logging to hotplug2-common.rules and it does log the relevant events. May 13 22:03:37 OpenWrt daemon.info hotplug: name=sda, path=/devices/platform/ar71xx-ehci/usb1/1-1/1-1:1.4/host4/target4:0:0/4:0:0:0/block/sda May 13 22:07:55 epointsystem: oh I know! May 13 22:08:11 epointsystem: it's because it's using the whole device not partitioned May 13 22:08:22 That's true. May 13 22:08:50 epointsystem: if you look at the mount code, it checks for sda1 vs sda May 13 22:09:08 Hmm. Let me see... May 13 22:12:17 Where is it? May 13 22:24:27 blkdev=`dirname $DEVPATH` May 13 22:24:28 if [ `basename $blkdev` != "block" ]; then May 13 22:25:12 sorry ...let me check again May 13 22:26:00 hmmm....ok, I lied, May 13 22:28:08 wait, no, I was right May 13 22:28:20 blkdev = .../block May 13 22:28:29 basename ..../block = block May 13 22:28:50 for partitions it is block/sda/sda1 May 13 22:29:36 this was necessary because otherwise the base device of a partitioned device tries to get mounted etc May 13 22:31:08 cshore: and when I rewrote the mount stuff I didn't change it because I thought it was a reasonable limitation (i.e. not worth the extra code to work around just so whole devices could be used instead of partitions) May 13 22:31:47 epointsystem: and I'm not sure what the fix would be in any case May 13 22:31:58 epointsystem: probably some nasty, vile hack May 13 22:33:02 Hmm. May 13 22:33:26 epointsystem: if we only had 2.6 it'd be okay but there are still 2.4 based systems May 13 22:33:28 Unfortunately, many microSD cards come in this fashion, without partitions, with the vfat filesystem directly on the drive. May 13 22:34:47 epointsystem: are you trying to allow random joe/jane user to use his/her microSD card? May 13 22:34:57 epointsystem: if not, just repartition May 13 22:36:33 epointsystem: otherwise you'll have to come up with a patch that doesn't break 2.4 or 2.6 (cause I have no idea) May 13 22:37:43 Well, tomorrow I'll discuss it with some co-workers. It's over midnight in my timezone and I am tired. But thank you very much for helping me understand the problem! I really appreciate it. May 13 22:37:49 epointsystem: maybe check if blkid shows it? May 13 22:38:12 epointsystem: busybox blkid is built in now May 13 22:38:25 epointsystem: ok I lied, I might have an idea for a patch May 13 22:38:26 Yeah: /dev/sda: UUID="CDF2-6BE2" May 13 22:39:32 I think the thing to do is instead of checking of blkdev != block, check that it is in blkid output May 13 22:39:57 Makes sense. May 13 22:40:56 epointsystem: well I'll let you sleep on it...I've got to get some paid work done :) May 13 22:43:53 Thanks once again. Good night. :-) May 14 01:41:30 build #44 of ar7 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar7/builds/44 **** ENDING LOGGING AT Fri May 14 02:59:56 2010