**** BEGIN LOGGING AT Fri Dec 04 03:00:42 2015 Dec 04 03:58:26 build #164 of ramips.mt7621 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7621/builds/164 **** BEGIN LOGGING AT Fri Dec 04 12:26:17 2015 Dec 04 12:32:43 jow r47754 trunk/target/linux/brcm63xx/base-files/etc/ (5 files in 3 dirs) * brcm63xx: switch from uci-defaults to board.d Dec 04 12:32:52 jow r47755 trunk/target/linux/ (5 files in 3 dirs) * lantiq: switch from uci-defaults to board.d Dec 04 13:52:39 build #136 of ramips.mt7620 is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7620/builds/136 Dec 04 13:56:03 build #160 of pxa is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/160 Dec 04 14:15:20 jow r47756 trunk/target/linux/ath25/ base-files/etc/uci-defaults base-files/etc/board.d/01_leds base-files/etc/board.d base-files/etc/board.d/02_network * ath25: switch from uci-defaults to board.d Dec 04 14:28:09 jow r47757 trunk/target/linux/ ar7/base-files/etc/board.d/02_network ar7/base-files/etc/uci-defaults ar7/base-files/etc/config ar7/base-files/etc/board.d * ar7: switch from uci-defaults to board.d Dec 04 14:33:00 jow r47758 trunk/target/linux/ sunxi/base-files/etc/uci-defaults sunxi/base-files/etc/board.d/02_network sunxi/base-files/etc/board.d * sunxi: switch from uci-defaults to board.d Dec 04 14:39:22 jow r47759 trunk/target/linux/orion/base-files/ etc/board.d/02_network etc/board.d etc/uci-defaults * orion: convert initial network configuration to board.d Dec 04 14:39:28 jow r47760 trunk/target/linux/sunxi/base-files/etc/board.d/02_network * sunxi: fix board.d permissions Dec 04 15:02:21 jow r47761 trunk/target/linux/ (5 files in 2 dirs) * mcs814x: switch from uci-defaults to board.d Dec 04 15:02:42 jow r47762 trunk/package/base-files/ files/lib/functions/uci-defaults-new.sh files/bin/config_generate * board.d: add support for gpio_switch sections Dec 04 15:02:50 jow r47763 trunk/target/linux/ar71xx/base-files/etc/ uci-defaults/01_gpio-switches board.d/03_gpio_switches * ar71xx: convert gpio switch config to board.d Dec 04 16:36:05 build #132 of au1000 is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/132 Dec 04 16:42:48 build #156 of ppc40x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/156 Dec 04 16:49:01 build #161 of imx6 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/imx6/builds/161 Dec 04 16:50:12 build #132 of brcm47xx.mips74k is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx.mips74k/builds/132 Dec 04 17:27:31 http://nbd.name/gitweb.cgi?p=luci2/procd.git;a=blob;f=plug/udevtrigger.c;h=3eba19a681e2edee1989dfc4fc35ca85f2f9d1f6;hb=HEAD Dec 04 17:27:31 line 165-167 -- if anyone knows why checking for dev, now.. ? Dec 04 17:28:17 Igel: hm, i think it was written that way because initially the code was used only to populate /dev Dec 04 17:28:31 what issue are you looking into? Dec 04 17:28:49 usb devices residing on an ftdi chip, Dec 04 17:29:13 where they are not being triggered by hotplug, because no /dev Dec 04 17:29:19 but.. they do have uevent Dec 04 17:29:46 not even when you reaatach them? Dec 04 17:29:54 iirtc udevtrigger is only used for cold-plugging Dec 04 17:30:05 i think also, i was able to do coldplug to get them to work, Dec 04 17:30:23 not even when reattached, the interfaces are not devs, evidently.. Dec 04 17:31:08 jow is right, this code has nothing to do with hotplugging Dec 04 17:32:13 interfaces come up on coldplug -- but .. may be a race condiiton? i would want to run it 1 time not twice Dec 04 17:33:45 going to recompile / test .. looking at doing (another?) coldplug at init.d/boot if this is not a fix. Dec 04 17:34:28 the check tho, on those lines is new, do you know what it was related to> just curious .. bout to grab some pizza, i'll peak back later thanks all, keep kicking ass.. etc.. :) Dec 04 17:43:35 nbd r47764 trunk/ (6 files in 6 dirs) Dec 04 17:43:35 lantiq: move esi calls to dsl_cpe_control scripts to fix ordering wrt. loading vr9 drivers Dec 04 17:44:08 nbd r47765 trunk/package/network/ utils/linux-atm/files/atm.hotplug utils/linux-atm/Makefile utils/linux-atm/files/br2684ctl * br2684ctl: convert init script to procd, add hotplug/reload support Dec 04 17:44:13 nbd r47766 trunk/target/linux/ lantiq/patches-4.1/0025-NET-MIPS-lantiq-adds-xrx200-net.patch lantiq/patches-4.1/0100-lantiq-xrx200-enable-remove-crc.patch * lantiq: clean up the xrx200 ethernet driver and fix stability issues Dec 04 18:35:21 nbd r47767 trunk/target/linux/lantiq/patches-4.1/0033-SPI-MIPS-lantiq-adds-spi-xway.patch * lantiq: fix xway spi driver timeout issues Dec 04 18:35:30 nbd r47768 trunk/target/linux/lantiq/dts/TDW89X0.dtsi * lantiq: re-enable spi-xway for TD-W89X0 now that it is fixed Dec 04 18:38:43 nbd: does that commit relate in any way to http://patchwork.ozlabs.org/patch/541967/ ? Dec 04 18:40:40 separate issue Dec 04 18:41:38 nbd: any chance you have one of the affected fritzboxen to try it? Dec 04 18:41:57 affected fritzboxen? Dec 04 18:42:01 which ones are those? Dec 04 18:42:05 and affected by what? Dec 04 18:43:11 FRITZ3370.dts Dec 04 18:43:43 by the spi-max-frequency Dec 04 18:44:25 nbd: ^ Dec 04 18:45:01 i may have that one somewhere Dec 04 18:46:05 you can post results to that ML thread Dec 04 18:48:18 add the edit to the relevant dts for the test Dec 04 18:48:36 nbd: ^ Dec 04 18:50:37 at the moment i care more about fixing other stuff Dec 04 18:50:44 so not sure if i'll get around to it Dec 04 19:00:00 ok it worked ndb, if people interface with usb interfaces over ftdi, like with libusb, they need to coldplug or echo into uevent.. so my change was for non-hotplug usb interfaces Dec 04 20:17:56 build #164 of ramips.mt7628 is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7628/builds/164 Dec 04 20:24:06 build #132 of ramips.rt305x is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.rt305x/builds/132 Dec 04 20:26:25 nbd r47769 trunk/package/network/config/ltq-vdsl-app/Makefile * lantiq: debloat the ltq-vdsl-app binary Dec 04 20:26:30 nbd r47770 trunk/target/linux/lantiq/patches-4.1/0033-SPI-MIPS-lantiq-adds-spi-xway.patch Dec 04 20:26:30 lantiq: fix a race condition in the SPI driver leading to rx FIFO overflows (and subsequent timeouts) Dec 04 20:26:41 ^^ that one was a tough nut to crack Dec 04 20:27:42 build #139 of avr32 is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/139 Dec 04 22:20:16 * russell-- has a kernel panic on x86_64 pcengines APU with r47681: http://pastebin.com/fnAm3dse Dec 04 22:39:50 nbd r47771 trunk/package/kernel/ (96 files in 2 dirs) * mac80211: update to 2015-12-03 Dec 04 22:40:46 nbd r47772 trunk/package/kernel/mac80211/Makefile * mac80211: enable build with linux 4.4 Dec 04 22:41:25 russell--: hm, probably a corrupted rcu_head data structure Dec 04 22:41:59 unfortunately it's probably impossible to tell from that log which module or part of the kernel messed up Dec 04 22:53:35 question, polarssl, why does it use shared mbedtls instead of static one? in theory it's just used by polarssl, no? Dec 04 22:54:29 considering that mbedtls is just the new name of polarssl, i have trouble understanding your question Dec 04 22:54:37 build #143 of xburst is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/143 Dec 04 22:56:29 https://github.com/openwrt/openwrt/blob/master/package/libs/polarssl/Makefile#L56 activates https://github.com/ARMmbed/mbedtls/blob/development/library/CMakeLists.txt#L2 Dec 04 22:58:21 so? Dec 04 22:58:34 polarssl/mbedtls is a library Dec 04 22:58:39 one that is used by several things Dec 04 22:58:49 why should we make it a static lib? Dec 04 23:00:23 nbd: well, my problem really is that for some reason, openwrt detect through readelf that I have a missing dependency Dec 04 23:00:30 when the library has been renamed Dec 04 23:00:41 although it is linked Dec 04 23:01:01 > 0x0000000e (SONAME) Library soname: [libmbedtls.so.9] Dec 04 23:02:35 does your package have a dependency on libpolarssl? Dec 04 23:02:42 in the package makefile Dec 04 23:03:06 ok, I think just renaming the .so wont work in this case. To be able to have my own version of the library I will have to patch CMakeLists.txt to make the output library name the one I need to openwrt detect correctly the library Dec 04 23:03:16 nbd: sure, but I have repackaged libpolarssl as libmpolarssl Dec 04 23:03:23 erm... why? Dec 04 23:03:46 nbd: because I am packaging every library I use in the target system I haven't control in to avoid issues Dec 04 23:04:03 you're insane, in case you hadn't worked that out yet ;) Dec 04 23:04:30 and you're going to keep battling those issues the longer you stay on AA. Dec 04 23:04:41 you're goign to have a massssive pile of patches and backports Dec 04 23:04:49 karlp: I have a parallel ubus with uhttpd, ubox, uci Dec 04 23:05:20 karlp: not really, until I found libpolarssl, it was enough to copy .so files with the name I wanted and it would work Dec 04 23:05:26 but for some reason in this case isn't working Dec 04 23:05:33 also, I created the ipkgmerger I talked about Dec 04 23:05:54 have fun man, it's your BBQ. Dec 04 23:05:55 and seems to be working Dec 04 23:05:57 ipkgmerger? Dec 04 23:06:10 nbd: to provide software in one ipk instead of 9 Dec 04 23:06:24 heh Dec 04 23:06:59 yeah, it basically untars all the control/data tarballs, keeps tracks on their external/internal dependencies and assembles a unique control file Dec 04 23:07:16 in the order they would be installed to avoid dependency issues Dec 04 23:07:38 I am hoping to be able to contribute it one day of these, but I need permission from my employer :( Dec 04 23:07:57 is a simple python script of ~400 lines Dec 04 23:08:21 three words just popped into my head: whiskey tango foxtrot Dec 04 23:08:27 hahaa Dec 04 23:08:51 yeah, in my case is: clown, circus, juggler Dec 04 23:16:14 the amount of time you've spent you'd be easily moved forward to current now... Dec 04 23:20:21 karlp: my company doesn't create the images, we are literally hacking the images Dec 04 23:20:56 left hand not talking to the right hand then? Dec 04 23:21:10 karlp: there are no hands I can talk to xD Dec 04 23:21:43 It's just me with a closed module we can install ipks in Dec 04 23:22:55 could just build an up to date openwrt and use its rootfs to generate a package you can chroot into ;) Dec 04 23:23:47 that way you have a cleaner split between the insane parts (base system) and the slightly more sane parts (the overlay) ;) Dec 04 23:24:25 btw, thinking on openwrt build system (trunk), when you build a package, the root it has access to, is just a sum up of the packages listed as dependencies? I mean, is a clean root with the dependencies? Because that way we could maybe have cleaner dependencies specified Dec 04 23:25:05 nbd: they control the deployment of the images xD Dec 04 23:25:58 what root are you talking about that a package would have access to? Dec 04 23:26:40 nbd: on the compilation process, if you specify that X depends on Y and Z, the makefile system will make sure Y and Z are there, but does it make that just Y and Z are there? Dec 04 23:27:25 what do you mean by 'just Y and Z are there'? Dec 04 23:27:57 when you talk abstract, it gets cryptic and weird Dec 04 23:28:03 how about you describe some examples Dec 04 23:29:21 ok, so I have found that in some cases, when you run into a magical -j number, you find out that a dependency is missing, just because usually people use smaller numbers, and package Y, which is a missing dependency of X, happens to be compiled before Dec 04 23:30:01 so, my idea was to compile X with just access to a rootpath where Y and Z are installed Dec 04 23:30:43 the build system already catches missing dependencies on shared libraries without that Dec 04 23:31:04 that way, even if it happens that Y is compiled/installed before X, X would fail because Y isn't in the root path Dec 04 23:31:07 since it does readelf on the binaries and looks at files describing which package provides which libraries Dec 04 23:31:09 nbd: that's the readelf one Dec 04 23:31:16 sure Dec 04 23:31:30 but what if the dependency isn't a library? Dec 04 23:31:38 what if it's just an script it depends on? Dec 04 23:32:04 you mean depends on at build time? Dec 04 23:32:14 not build time, but runtime Dec 04 23:32:37 well, maybe build time to Dec 04 23:32:49 how is the build system supposed to check for that? Dec 04 23:32:50 at the end, runtime can't be expected and would be too comples Dec 04 23:33:45 nbd: I was just trying to think that if you could actually cache packages, and extract them in a clean rootfs, you would be able to speedup builds, because those packages could be reused Dec 04 23:34:10 thinking on DevInstall packages more preciselly Dec 04 23:35:30 at the moment, i don't see much point in that Dec 04 23:35:31 I mean, the build system to change so that a dependency means that DevInstall of packages X and Y is done, rather than X and Y being on readelf path Dec 04 23:36:15 copying stuff together on demand also takes time and disk space Dec 04 23:39:17 and if combined with overlay filesystems? Dec 04 23:39:39 erm.... what? Dec 04 23:40:08 how is that supposed to work? Dec 04 23:40:51 so lets say you have all staging_dir/targetXXX/dev-packages/{X,Y,Z} Dec 04 23:41:38 those dirs have devinstalls of packages X, Y, Z Dec 04 23:42:11 you can instruct the kernel to create a filesystem with Y and Z overlayed Dec 04 23:42:23 and act as a root for the compilation of X Dec 04 23:42:43 that way you avoid the copies Dec 04 23:42:55 and you just have to delete the namespace Dec 04 23:43:46 probably an overkill Dec 04 23:43:50 solutions looking for problems mand Dec 04 23:44:49 karlp: I heard someone complain about how long builds were Dec 04 23:45:55 anyway, yeah, I will leave it ther Dec 04 23:48:11 that makes me wonder... does your employer provide you with a stash of drugs or do you buy them yourself? ;) Dec 04 23:50:13 nbd: I am not in drugs... If you think that of me like this without them... It would probably horrible me with them Dec 04 23:50:22 heh Dec 04 23:50:27 or you'd see reason :) Dec 04 23:50:34 haha probably Dec 05 00:03:05 OMG! I totally overkilled. I was cleaning the polarssl package instead of the one that was giving me the dependency problem, so the compilation step was always skipping, and was always checking against an old binary :/ **** ENDING LOGGING AT Sat Dec 05 02:59:58 2015