**** BEGIN LOGGING AT Wed Feb 17 02:59:59 2016 Feb 17 06:25:32 mamarley: please work on it, so far no developer expressed interest Feb 17 07:25:10 mamarley: afaik some ppl are using glibc on non octeon (self build images) - afaik the glibc 2.22 patches were not by someone who used octeon (since that platform is quite exotic) Feb 17 08:44:41 jow r48733 trunk/include/download.mk * download.mk: fix packed checkout mirroring support Feb 17 08:44:46 jow r48734 trunk/toolchain/glibc/Config.in * toolchain: mark eglibc 2.19 broken Feb 17 08:44:52 jow r48735 trunk/ toolchain/glibc/patches/2.22/200-add-dl-search-paths.patch toolchain/glibc/common.mk toolchain/glibc/patches/2.21/200-add-dl-search-paths.patch toolchain/glibc/Config.version * toolchain: use latest glibc 2.21 and 2.22 revisions Feb 17 11:06:36 jow_laptop: Thanks and sorry. I was looking at it and trying to figure out the right git revision to use, but it got really late and I had to get some sleep. Feb 17 11:06:57 I will test and backport it to CC. Feb 17 11:12:49 mamarley: np **** BEGIN LOGGING AT Wed Feb 17 12:00:09 2016 Feb 17 13:23:54 anyone know what http://git.openwrt.org/?p=project/mountd.git;a=blob;f=mount.c;h=88920407490a97f71f668e320c918090e936cd87;hb=HEAD#l389 is meant to be trying to do? it seems to just mangle the serial to add lots of 6s, but I can't understand why one would ever want that? Feb 17 13:24:55 it means that the uci options looked up have no relation to the serial of the disc itself. Feb 17 13:27:50 looks like a simple checksum calculation Feb 17 13:27:58 s/checksum/fingerprint/ Feb 17 13:29:16 it calculates a uuid from a serial string of arbritary length Feb 17 13:30:19 for what purpose? why would I want to use a uci mount option name that's not the serial of the disk and not shown anywhere? Feb 17 13:30:35 I can understand _maybe_ if ther was no serial found, rather than just having 000000 but when there _is_ a serial? Feb 17 13:31:27 it synthesizes serial numbers Feb 17 13:31:36 or rather uuids from serial numbers Feb 17 13:32:28 maybe for some /dev/disk/by-uuid/ kind of thing Feb 17 13:33:37 mountd also keeps an internal state database; it is possible that it uses the uuids there to uniquely identify block devices Feb 17 13:33:58 at least one spot in the code uses the fabricated uuid as uci section id Feb 17 13:34:15 yeah, that's what I was trying to debug, Feb 17 13:34:26 what section id I was meant to use, and where on earth it was coming from. Feb 17 13:36:41 given that's it's not a real uuid anyway, what advantage would this even have? two devices with the same serial with have the same "pseudo discid" anyway, so I don't see what this is gaining. Feb 17 13:40:24 a uniformely formatted uuid Feb 17 13:40:37 only consisting of valid hexadecimal digits Feb 17 13:40:49 not some "random chine backyard usb_disk 122ööäü" Feb 17 13:41:17 the serial it uses as input to calculate the uuid from can come from various sources Feb 17 13:41:24 embedded usb identifiers, labels, sysfs files Feb 17 13:41:30 it may contain spaces, slashes etc. Feb 17 13:42:45 the pseudo ssid ensures a) a fixed length b) no special chars Feb 17 13:42:58 its like running md5sum over an arbritary input, just simpler Feb 17 13:44:32 ok, that's a good enough reason :) Feb 17 16:38:57 hi! "logread" is always empty on current trunk on Raspberry Pi 2. any ideas? Feb 17 17:40:37 arm SDK from 15.05 can not build a hellworld with -static flag, mips SDK is fine, what gives? Feb 17 17:42:19 (.ARM.exidx+0x0): undefined reference to `__aeabi_unwind_cpp_pr1' Feb 17 17:49:30 luka r48736 trunk/target/linux/ (8 files in 7 dirs) * mvebu: add SolidRun ClearFog A1 support Feb 17 17:51:48 luka r48737 branches/chaos_calmer/target/linux/mvebu/ (8 files in 6 dirs) * mvebu: add SolidRun ClearFog A1 support Feb 17 17:56:21 ausjke: try -lgcc_eh Feb 17 18:06:28 jow_laptop: tried that, in fact tried -lgcc_eh, -lgcc_initial -lgcc_(forgot), that contains the unwind string Feb 17 18:06:50 those three are the only static libraries containing the missing string Feb 17 18:07:58 libgcc_pic.a, libgcc_eh.a, libgcc_initial.a Feb 17 18:08:21 trying to compile static with androkd ndk now Feb 17 18:08:50 I think my block mount problems are that when the sdcard is removed from teh card reader, I get a block action "change" which "block" ignores, it only knows how to deal with "remove" Feb 17 18:08:59 to help an android apps developer to run a static binary on his phone for quick tests Feb 17 18:09:52 but, openwrt SDK arm really should be able to build a static helloworld Feb 17 18:20:34 ausjke: canont really help for that, "arm" for me always was the thing that is different from the rest and rather obscure Feb 17 18:21:56 jow_laptop: thanks. I probably will submit a bug report. tried android-ndk and that builds fine, good for now. Feb 17 18:22:56 ausjke: I believe there is an open bug already and noone cared Feb 17 18:23:38 https://dev.openwrt.org/ticket/19588 Feb 17 18:24:29 I will have a look at that Feb 17 19:30:37 jow_laptop: https://wiki.linaro.org/KenWerner/Sandbox/libunwind Feb 17 19:30:54 looks like ARM just need a flag "-funwind-tables" Feb 17 19:31:19 i'm going to build a SDK and test it Feb 17 20:09:34 hi there, I would like to submit an openwrt config diff for a specific target (Raspberry PI) Feb 17 20:10:30 how can I do that? I have generated the diff using /scripts/diffconfig.sh > diffconfig Feb 17 20:11:18 should I send the diff to the mailing list.. or what? Feb 17 20:11:45 gnuton: https://dev.openwrt.org/wiki/SubmittingPatches Feb 17 20:17:34 wigyori, it doesn t say anything about .config files Feb 17 20:18:02 why do you want to submit a .config file anyway? Feb 17 20:18:29 we don't have configs in the repo Feb 17 20:18:32 because it enables some kernel modules which makes lxc packages working for raspberry Feb 17 20:18:35 non-kernel ones that is Feb 17 20:18:59 currently all lxc packages are useless Feb 17 20:19:16 they require you to build yourself with certain kernel features enabled Feb 17 20:19:32 sounds like a joke Feb 17 20:19:50 so openwrt provides half broken packages and no lxc kernel support Feb 17 20:19:59 basically nothing works Feb 17 20:20:32 those packages are provided by community feeds Feb 17 20:24:21 I ll try to get it fixed in any case... Feb 17 20:24:56 RPI are powerfull enough to get few KB of fatter kernels **** ENDING LOGGING AT Thu Feb 18 02:59:59 2016