**** BEGIN LOGGING AT Mon Sep 28 02:59:58 2015 Sep 28 06:06:06 build #100 of ppc40x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/100 Sep 28 06:46:25 build #91 of brcm47xx.mips74k is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx.mips74k/builds/91 Sep 28 08:10:41 plntyk: hi! we talked about booting malta in qemu using the harddrive image. I've tried to compile little endian (to be same as the wiki example), and I get the same error (/dev/sda is not present when booting the non-initramfs kernel image; root fails to mount). any ideas? I'm running "QEMU emulator version 2.4.0.1" Sep 28 08:11:04 plntyk: I'd be happy to help debug this if I knew what to check Sep 28 08:14:31 attila_lendvai: hm... its not in CC (45974 CC branch), changeset for ext4 https://dev.openwrt.org/changeset/46269 Sep 28 08:15:14 weren't you using CC? Sep 28 08:15:20 plntyk: isn't it enough for me on CC to turn on those in the patch in menuconfig? Sep 28 08:16:18 attila_lendvai: no - because you need a kernel that supports the sda device and the ext4 file system driver in the kernel image Sep 28 08:16:45 maybe I should move on to trunk? this thing won't be in production for at least 3 more months... maybe I should move with the bleeding edge Sep 28 08:17:35 oh, I see. there's a separate kernel config that I cannot see in make menuconfig! I'll just backport that change then... Sep 28 08:17:37 backporting the patch should be simple enough too - its just integrating the drivers into the kernel so that it can find the hw and rootfs that you provide Sep 28 08:18:07 * attila_lendvai ponders whether to backport, or to escape forward... Sep 28 08:18:26 the initramfs has the "filesystem" integrated so its different from a "kernel +rootfs" that you use with qemu and non initramfs kernel Sep 28 08:19:01 yeah, that's what I was guessing when I looked at the filename and the .elf extension Sep 28 08:20:55 the next release after CC will probably be a year down the road from the release (as opposed to 15.05), right? Sep 28 08:42:05 hopefully :) Sep 28 09:10:24 rmilecki r47064 trunk/package/network/services/hostapd/patches/600-ubus_support.patch * hostapd: check for banned client on association event Sep 28 09:41:08 rmilecki r47065 branches/chaos_calmer/package/network/services/hostapd/patches/600-ubus_support.patch * hostapd: check for banned client on association event Sep 28 09:42:00 nbd r47066 trunk/package/base-files/Makefile * base-files: add a dependency on fstools Sep 28 10:08:35 build #93 of ramips.mt7620 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7620/builds/93 Sep 28 11:01:17 wpan-tools may be missing some compile time dependency on libnl? I get this on HEAD when compiling it: "netlink/genl/genl.h: No such file or directory" Sep 28 11:05:29 lots of libnl tweakage in progress, there's a bunch of issues on github Sep 28 11:05:53 is there a moving tag that marks a stable snapshot? Sep 28 11:07:42 probably not, I cannot find anything in the svn or git repos Sep 28 11:10:13 yeah, the release branches :) Sep 28 11:10:43 oh well... the release branches are too old, and I chose not to backport but to run forward. may have been a mistake... Sep 28 11:11:11 CC is too old? Sep 28 11:11:26 I needed some patch not in CC Sep 28 11:15:34 * attila_lendvai picks that commit, reset --hard, and restarts the build Sep 28 11:17:13 nbd: uml does not seem to build for me in trunk, even with rm .config ; make menuconfig, select uml, and build from scratch. the 4.1 upgrade is what broke it (it blows up somewhere in depths of __uml_setup macro in arch/um/kernel/config.c); the previous commit with 3.8.21 kernel builds fine. Sep 28 11:17:43 that is, git commit 07efcbd1ec2858f55e9927b98578a3bc06e6f796 is the last one that works (svn 47059) Sep 28 11:17:45 idli: please show me a build log Sep 28 11:18:19 it worked just fine for me, i tested 32 bit and 64 bit Sep 28 11:18:26 nbd: http://pastebin.com/Y6SrApt9 Sep 28 11:18:32 that's where it blows up with -j1 V=s Sep 28 11:19:10 64bit platform, default .config (except chose uml target) Sep 28 11:20:57 what distro are you using? Sep 28 11:21:19 debian 8 Sep 28 11:23:01 hm, i'm using debian testing Sep 28 11:23:22 will take another look to see if i can reproduce it somehow Sep 28 11:23:24 or figure out what causes it Sep 28 11:24:21 it is rather weird, it is in a part that is not patched by OpenWrt UML target Sep 28 11:24:34 and I have used similar (but not quite same) UML kernel built natively on that Sep 28 11:24:58 different compiler is the only difference I would hazard, platform default is 4.9.2 Sep 28 11:26:38 hm, wonder if my clean wasn't quite clean enough, guess I will check Sep 28 11:28:17 4.1.6 built 'natively' for UML fine, I did now make dirclean + rm .config + make menuconfig + select UML + make -j 9 Sep 28 11:28:40 we'll see what the result is in 15-20min or so Sep 28 11:41:12 idli: hm, now i see the error too Sep 28 11:41:16 not sure why it didn't happen before Sep 28 11:42:10 ah, good, it wasn't just in my head :) Sep 28 11:42:22 yep, second build from scratch resulted in same Sep 28 11:57:03 found the issue Sep 28 11:57:05 need to backport another commit Sep 28 11:57:10 testing a clean build right now Sep 28 12:23:56 nbd r47067 trunk/target/linux/uml/patches-4.1/ 001-um-Stop-abusing-__KERNEL__.patch 002-um-Remove-copy-paste-code-from-init.h.patch 001-um-Remove-copy-paste-code-from-init.h.patch * uml: backport another upstream patch to fix compile errors Sep 28 12:24:02 idli: should be fixed now Sep 28 12:24:45 nbd: thanks :) Sep 28 12:48:19 rmilecki r47068 branches/barrier_breaker/package/network/services/hostapd/patches/600-ubus_support.patch * hostapd: check for banned client on association event Sep 28 13:14:31 plntyk: got malta running in qemu with master. thanks for all your help! Sep 28 13:15:10 * attila_lendvai goes to the wiki and edits that page to make it less painful for others Sep 28 13:42:13 attila_lendvai: what else was missing? kernel_menuconfig to turn on those options, anything else? Sep 28 13:43:21 karlp: nothing was really missing. what *I* missed was that CC doesn't have the require patch (or kernel_menuconfig, which is also new to me) Sep 28 13:43:45 and unfortunately the failure mode didn't contain much hints Sep 28 13:45:12 error messages are only useful after you know what you were missing :( Sep 28 13:47:59 that's not true generally, but certainly true in linux hacking world. e.g. I have spent years working in common lisp environments where I have rarely spent time debugging in the sense one debugs in C and linux. 99% of the time one simple gets a backtrace and looks around in the context for broken assumptions. Sep 28 13:57:42 FTR, saved changes here: http://wiki.openwrt.org/doc/howto/qemu#openwrt_in_qemu_mips Sep 28 15:33:10 there's no m4 package or I just cannot find it? Sep 28 15:33:58 MMC_DEBUG in kernel_menuconfig should work for sd cards on usb card readers too right? Sep 28 15:35:10 "block info" will detect cards ok, Sep 28 15:35:15 attila_lendvai: you are correct no m4 package Sep 28 15:35:22 but I don't get any events on card insertion/removal, Sep 28 15:36:08 * attila_lendvai wonders if he's stretching the use-cases by running ./configure scripts and compiling stuff with gcc in a qemu user space environment... :) Sep 28 15:38:30 attila_lendvai: probably not - you could test gcc code generation for bugs this way (cross / "native") Sep 28 15:45:21 karlp: check documentation/code when that symbol was introduces - maybe you need to set some module parameter/debugfs entry ... Sep 28 15:58:42 * attila_lendvai is about to dl and compile gnu make in the qemu env and is wondering whether he's lost in the forest Sep 28 16:03:24 plntyk: I guess mmc debug might be the wrong symbol, as it's an sd slot on a usb device, not a native mmc/sd slot. Sep 28 16:06:12 karlp: maybe - http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/mmc/host/Kconfig lists MMC_USHC or MMC_REALTEK_USB - so probably depends what reader / driver / symbol is used Sep 28 16:09:08 hrm, not sure it's that, that's turnning on the hardware support itself, Sep 28 16:09:13 and the card reader _works_ Sep 28 16:09:21 I can mount and read/write no problems Sep 28 16:09:33 I'm just not getting card insertion/removal events as I'd expect. Sep 28 16:09:41 and not sure if it's hardware or software. Sep 28 16:10:25 we've followed the eval board schematic for teh hub+sd combo chip, but we don't have an eeprom connected for non-default configs. Sep 28 16:10:25 hrm, gnu make -v segfaults, --help works. is that surprising news, or not too unexpected given the context? Sep 28 16:10:59 build #94 of avr32 is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/94 Sep 28 16:42:27 plntyk: I can't even get dynamic debug to work. I have CONFIG_KERNEL_DYNAMIC_DEBUG=y in regular menuconfig, and the /sys/kenrnel/debug/dynamic_debug/control file exists and is full of entries, but turning anything on there does nothing at all. Sep 28 20:25:10 build #91 of mpc83xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/91 Sep 28 20:27:59 build #91 of ep93xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/91 Sep 28 21:59:37 nbd: It looks like commit 89c8104625aa80a8c874a78eeae8d553415173db in procd is causing a FTBFS because the return value of the symlink() function is ignored. Sep 28 23:35:05 Any good place to start reading docs or code to grok luci internals? Sep 29 00:43:03 build #95 of arm64 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/arm64/builds/95 Sep 29 00:57:12 build #97 of netlogic is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/netlogic/builds/97 **** ENDING LOGGING AT Tue Sep 29 02:59:58 2015