**** BEGIN LOGGING AT Tue Jan 19 02:59:58 2016 Jan 19 05:00:31 jow or ndb around? Jan 19 05:00:36 have a question with this sysupgrade stuff Jan 19 05:54:38 http://pastebin.com/TAh5a1zW haven't tested it yet Jan 19 05:54:55 but should give an idea of what im working on to write directly to each partition without blowing out the mbr Jan 19 06:27:03 nyt: probably not yet awake since they are in CET and its still just 7:30am local time Jan 19 10:00:55 hmmm bot broken ? Jan 19 10:01:32 ah no, patches accepted but still in someone's local tree I guess :) Jan 19 10:06:59 nbd r48343 trunk/package/network/services/uhttpd/Makefile * uhttpd: add option for mbedtls Jan 19 10:07:04 nbd r48344 trunk/package/ network/services/hostapd/files/hostapd-full.config network/services/hostapd/files/wpa_supplicant-full.config * hostap/wpa_supplicant: enable EAP-FAST in -full builds Jan 19 10:07:09 nbd r48345 trunk/package/network/services/hostapd/files/netifd.sh * wpa_supplicant: add support for EAP-TLS phase2 Jan 19 10:07:14 nbd r48346 trunk/target/linux/generic/config-4.4 * kernel: add missing symbol for 4.4 Jan 19 10:07:24 nbd r48347 trunk/package/kernel/linux/modules/crypto.mk * kernel/modules: add kmod-crypto-hw-ccp Jan 19 10:07:33 nbd r48348 trunk/target/linux/generic/patches-4.4/150-crypto-ccp-add-hash-state-import-and-export-support.patch * kernel: add patch to fix the ccp-crypto module Jan 19 10:07:48 stintel: ;) Jan 19 10:10:25 morning :) Jan 19 10:10:47 curious to see strongswan performance on my apu2 Jan 19 10:15:45 blogic r48349 trunk/target/linux/ (14 files in 3 dirs) * at91: add 4.4 support Jan 19 10:15:55 blogic r48350 trunk/target/linux/at91/modules.mk * at91: add package for high speed usb device Jan 19 10:16:02 blogic r48351 trunk/target/linux/at91/image/Makefile * at91: remove SAMA5D3 from Default image build. Jan 19 10:16:07 blogic r48352 trunk/target/linux/at91/modules.mk * at91: fix typo in usb kernel module makefile Jan 19 10:16:15 blogic r48353 trunk/target/linux/at91/Makefile * at91: move to 4.4 kernel Jan 19 10:16:19 blogic r48354 trunk/target/linux/ at91/config-3.18 at91/patches-3.18 * at91: drop v3.18 Jan 19 10:16:27 blogic r48355 trunk/target/linux/ramips/image/Makefile * ramips: HiLink HLK-RM04 image name typo (RM02 -> RM04) Jan 19 10:16:44 blogic r48356 trunk/target/linux/ (12 files in 9 dirs) * ar71xx: add support for Netgear WPN824N Jan 19 10:16:52 blogic r48357 trunk/target/linux/ar71xx/files/arch/mips/ath79/mach-dlan-pro-1200-ac.c * ar71xx: fix AR8337 switch initialization on dLAN pro 1200+ WiFi ac Jan 19 10:17:03 blogic r48358 trunk/target/linux/ar71xx/base-files/etc/board.d/02_network * ar71xx: board.d: fix switch defaults on dLAN pro 1200+ WiFi ac Jan 19 10:17:11 blogic r48359 trunk/target/linux/ramips/dts/ZBT-WG2626.dts * ramips: fix serial baudrate for final version of ZBT-WG2626 board Jan 19 10:17:17 blogic r48360 trunk/target/linux/ (7 files in 6 dirs) * ipq806x: Add support for Netgear D7800 Jan 19 10:17:28 blogic r48361 trunk/ (5 files in 4 dirs) * linuxptp: Add package Jan 19 10:20:57 blogic r48362 trunk/net * Revert "linuxptp: Add package" Jan 19 12:03:06 jow_laptop: what do you think of https://github.com/openwrt/packages/pull/2283/files now? I've cleaned up all of whatever was still pulling from my old feed now Jan 19 12:03:48 karlp: looks good to me Jan 19 12:16:05 that should have fixed things for the 1505 branch now too Jan 19 12:16:55 did you cherry-pick ? Jan 19 12:17:20 took the branch that had them and rebased it on for-1505 Jan 19 12:17:41 do you want to use a script that links them together with "back port of x" notes or something? Jan 19 12:20:22 no, I just wanted to know whether I can trigger a rebuild for 15.05 Jan 19 12:20:43 nbd, ping Jan 19 12:20:53 well, from john's mail it seems he had to wait on a few other packages too right? Jan 19 12:21:06 nitroshift: pong Jan 19 12:21:30 though, kinda hard for you know, _anyone_ in the "community" to have helped much given that no-one even knew that 1505.1 builds were already underway. Jan 19 12:21:37 nbd, hello :) just wanted to ask whether 15.05.1 will be out soon Jan 19 12:21:49 not sure. i think it's still being worked on Jan 19 12:21:54 i see Jan 19 12:21:56 thanks Jan 19 12:22:03 should hopefully be a matter of days Jan 19 12:22:06 not weeks Jan 19 12:22:11 nbd, there's a witi2 in the works Jan 19 12:22:14 interesting Jan 19 12:22:25 any details on that? Jan 19 12:22:34 still mt7621 based Jan 19 12:22:46 karlp: I keep building packages out of band for 15.05 using the published 15.05 release sdks Jan 19 12:23:04 karlp: its independant of the 15.05.1 release process Jan 19 12:23:05 but mqmaker will go through the process to certify it with ecc / fcc Jan 19 12:23:29 from what i gathered, current witi was just a market prospection Jan 19 12:23:41 nbd, pm please? Jan 19 12:23:45 jow_laptop: and that's great, I'm j ust saying that the total absence of communication about an upcoming point relase makes it hard for community members to help in anything meaningful. Jan 19 12:23:49 sure Jan 19 12:24:33 karlp: there is not much to help actually Jan 19 12:24:34 mosquitto 1.4.7 came out a month and a half ago, I would have pushed things a lot further forward on my plate of things to do if there was a release coming, rather than just slowly getting to it when it came up for me. Jan 19 12:25:52 karlp: and thats fine, one of the ideas about frequent releases or just point releases is that there is no need to forcibly cram everything into a given deadline just because there won't be another release in the next two years Jan 19 12:26:16 karlp: we have the ifnrastructure in place to do out of band package updates so its no issue to bump packages after a poitn release Jan 19 12:42:40 nbd r48363 trunk/toolchain/musl/common.mk * toolchain/musl: enable debug information Jan 19 12:48:57 btw. is anybody here aware of any remaining issues with binutils 2.25? Jan 19 12:49:06 i'd like to update to that soon Jan 19 12:49:26 update first, then ask again :) Jan 19 12:50:24 Are there any plans to upgrade to glibc 2.22? Jan 19 12:50:45 i'm fine with upgrading glibc if somebody sends patches for it Jan 19 12:50:52 i'll even merge them Jan 19 12:50:57 i just don't have any plans to work on it myself Jan 19 12:51:19 * mamarley might do that, if he can figure it out. Jan 19 12:51:20 the guy working on mips64 support for musl seems to have disappeared :( Jan 19 12:51:21 http://www.openwall.com/lists/musl/2015/10/27/7 Jan 19 12:51:26 and then nothing Jan 19 12:51:38 hm, too bad Jan 19 12:52:54 nbd: Did you see the thing I posted a few days ago about the save and restore masks for the qos-scripts iptables rules preventing ingress traffic prioritization from working? Jan 19 12:53:21 mamarley: where did you post it? Jan 19 12:53:29 In here. Let me see if I can find it. Jan 19 12:54:17 [15.01.16 12:59:05] nbd: When I went to get my qos-scripts patches ready to submit, I noticed https://github.com/openwrt-mirror/openwrt/commit/5e914350f80f985e8915996410e4e9bbd5e2e5ff, which seems to be the cause of one of my issues. With the connection mark split like that, I can't seem to get any of my rules to apply to ingress traffic. For example, if I configure it to classify all traffic from my guest network as "bulk" Jan 19 12:54:18 , the egress traffic is properly deprioritized if there is a heavy load. However, ingress traffic is not affected at all and the priority appears equal to the rest of the traffic. It looks to me like the save-mark rule saves half of the mark and the restore-mark rule restores the other (unsaved) half. Jan 19 12:57:12 hm Jan 19 12:57:14 interesting Jan 19 12:57:25 If I modify generate.sh to use 0xff for both the save and restore masks, it works just fine. Jan 19 12:59:28 Also, if I configure my rule as a "default" instead of a "classify", it works fine. Jan 19 13:01:59 how about if you keep the restore mask at 0x0f, but set the save mask to 0xff? Jan 19 13:02:29 that should be closer to the intention of that commit Jan 19 13:16:44 nbd: OK, I will try that. Jan 19 14:01:01 I browsed the wiki in and out but couldn't find any info on ubus internals Jan 19 14:01:17 usage patterns/idioms, concepts ... Jan 19 14:01:27 does such a thing exist, or is it all in the code? Jan 19 14:05:47 normally, someone has a need, gets told "use ubus" and if you're lucky, they write down what they learn in the wiki. Jan 19 14:06:06 and try and scour teh mailing list for snippets too Jan 19 14:41:21 yup, thought so. thanks, karlp Jan 19 15:15:31 nbd r48364 trunk/package/kernel/ linux/modules/fs.mk linux/modules/netsupport.mk * kernel: disable AFS support by default Jan 19 15:15:52 nbd r48365 branches/chaos_calmer/ package/kernel/linux/modules/fs.mk package/kernel/linux/modules/netsupport.mk * kernel: disable AFS support by default Jan 19 15:32:15 blogic r48366 trunk/package/system/procd/files/hotplug.json * procd: emit events for sound subsystem Jan 19 16:52:58 nbd r48367 trunk/target/linux/cns3xxx/config-4.4 * Revert "cns3xxx: re-enable CONFIG_CPU_SW_DOMAIN_PAN" Jan 19 17:40:10 So I was able to get openwrt trunk built with glibc 2.22, but when I try to boot it on an x86_64 VM, it gets stuck at "procd: - ubus -" without any error messages or other abnormal output. Is there any way to debug that? Jan 19 17:41:54 mamarley: try to raise the debug level Jan 19 17:42:40 jow_laptop: How to do that? Is there some kernel command argument or something?, because it never gets far enough for me to supply any input to the system itself. Jan 19 17:43:22 try passing init_debug=4 via grub Jan 19 17:46:16 jow_laptop: When I do that, I get an extra message asking me to press "f" for failsafe mode or 1-4 and enter for selecting a debug mode, but that is all. Jan 19 17:57:09 jow_laptop: It locks right before it would normally say "random: ubusd urandom read with xx bits of entropy available". Perhaps this has something to do with entropy? Jan 19 17:57:39 maybe the random syscall in uclibc is broken Jan 19 17:57:47 erm... brainfart Jan 19 17:58:12 maybe the random call in glibc works differently compared to musl Jan 19 17:58:21 maybe it uses blocking random reads vs. nonblocking ones Jan 19 17:58:27 It worked fine in glibc 2.21. Jan 19 17:58:57 maybe its something with the VM? that random thing in log is taking a while in musl too (afaik ~40 seconds) - so often the rootfs/welcome message is printed earlier Jan 19 17:59:26 bisect glibc :) Jan 19 17:59:41 plntyk: I doubt it. I have two VMs running side by side, one compiled with 2.21 and one with 2.22. Jan 19 17:59:55 what host/VM soft are you running on ? Jan 19 18:00:01 KVM Jan 19 18:00:44 isnt that qemu ? Jan 19 18:01:07 Yeah, the executable is "qemu-system-x86_64". Jan 19 18:01:33 can you post your 2.22 patch ? Jan 19 18:01:47 or is it just ver+md5sum ? Jan 19 18:02:47 plntyk: It isn't entirely ready yet. I manually added "libutil.so.1" to the libc.provides file and I would still need to figure out how to do that automagically. Beyond that, it is just adjusting the makefiles to download and compile the new version. Jan 19 18:04:43 ubusd reads /dev/urandom directly, so no library call involved Jan 19 18:04:58 I'd say that means that ubusd never starts Jan 19 18:05:28 mamarley: for testing I suggest to patch procd to set the "debug" variable unconditionally to 4 Jan 19 18:05:33 I can get it to boot up to a shell by passing init=/bin/ash to the kernel, if that might be helpful for debugging something. Jan 19 18:05:49 mamarley: it seems the debug level enabling via cmdline is broken (will take a look at it when I'm home) Jan 19 18:06:05 Sure, I will try that. Jan 19 18:06:30 it should report more Jan 19 18:06:37 at least whether ubus connect failed or not Jan 19 18:06:47 and some other random info like state changes etc. Jan 19 18:08:16 jow_laptop: If I select the failsafe boot option in grub, it gets stuck on "init: - preinit -" instead. I will try hardcoding the debug level now. Jan 19 18:09:11 hmm Jan 19 18:09:24 it should spawn a shell at this point Jan 19 18:09:46 mamarley: regarding libutil - look in package/libs/toolchain/Makefile there is a versioned CONFIG_GLIBC_USE_VERSION_2_21 line - so probably change that Jan 19 18:10:07 i looked at what changed in r44701 and found that Jan 19 18:11:34 plntyk: Ah, thanks! I already fixed one of those 2_21 conditionals, but I missed at least that one. Jan 19 18:20:49 Hmm, makefiles are fun. *heads off to read documentation on make conditionals* Jan 19 18:59:55 jow_laptop: plntyk: After fixing all the remaining 2_21 conditionals, it boots just fine. I will send a patch to the ML shortly. Thanks for the help! Jan 19 20:59:29 rmilecki r48368 trunk/target/linux/ (14 files) * bcm53xx: group iProc patches into patchsets they were sent in Jan 19 20:59:34 rmilecki r48369 trunk/target/linux/ bcm53xx/patches-4.1/034-PCI-iproc-Fix-BCMA-dependency-in-Kconfig.patch bcm53xx/patches-4.1/035-PCI-iproc-Allow-BCMA-bus-driver-to-be-built-as-modul.patch * bcm53xx: backport iProc Kconfig fixes Jan 19 20:59:51 rmilecki r48370 trunk/target/linux/ (16 files) * bcm53xx: move iProc patches that went mainline Jan 19 21:09:52 nbd r48371 trunk/ (6 files in 5 dirs) * kernel: modularize dma shared buffer code to eliminate kernel bloat Jan 19 21:09:57 nbd r48372 trunk/target/linux/ generic/patches-3.18/010-KEYS-Fix-keyring-ref-leak-in-join_session_keyring.patch generic/patches-4.3/010-KEYS-Fix-keyring-ref-leak-in-join_session_keyring.patch generic/patches-4.1/012-KEYS-Fix-keyring-ref-leak-in-join_session_keyring.patch generic/patches-4.4/010-KEYS-Fix-keyring-ref-leak-in-join_session_keyring.patch * Jan 19 21:09:59 OpenWrtkernel: fix keyring reference leak (CVE-2016-0728) Jan 19 21:10:40 nbd r48373 branches/chaos_calmer/target/linux/generic/patches-3.18/010-KEYS-Fix-keyring-ref-leak-in-join_session_keyring.patch * kernel: fix keyring reference leak (CVE-2016-0728) Jan 19 21:50:52 wigyori r48374 trunk/package/ boot/uboot-mxs/patches/100-gcc-5-fix.patch boot/uboot-mxs/Makefile boot/uboot-mxs/patches/001-add-i2se-duckbill.patch * packages: bump uboot-mxs to 2016.01, refresh patches Jan 19 21:54:04 rmilecki r48375 trunk/target/linux/ (20 files in 2 dirs) * bcm53xx: update iProc patches (V5 accepted in helgaas/pci.git next) Jan 19 22:04:24 wigyori r48376 trunk/target/linux/generic/config-4.4 * kernel: add missing symbols for 4.4 Jan 19 22:05:46 wigyori r48377 trunk/target/linux/ mxs/patches-4.4/100-crypto-mxsdcp-provide-importexport.patch mxs/patches-4.4 mxs/config-4.4 * mxs: initial 4.4 support Jan 19 22:37:56 mamarley: I hacked '-d 4' into initd/preinit.c's /sbin/procd invocations... Jan 19 22:39:05 also ran into a dropbear openpty glibc 2.22 issue Jan 19 22:41:59 nbd r48378 trunk/package/libs/uclient/Makefile * uclient: update to the latest version, fixes a cosmetic progress bar issue Jan 19 22:42:12 nbd r48379 trunk/package/libs/uclient/Makefile * uclient: install a symlink from uclient-fetch to wget into /bin Jan 19 22:42:17 nbd r48380 trunk/package/libs/ustream-ssl/Makefile Jan 19 22:42:18 ustream-ssl: update to the latest version, fixes connection with servers requiring DHE Jan 19 22:48:50 swalker: What issue was that? After I applied the fix suggested by plntyk, it seemed to work fine with glibc 2.22. Jan 19 22:49:18 nbd r48381 trunk/target/linux/lantiq/config-4.4 * lantiq: Enable the uImage mtd splitter Jan 19 22:54:02 rmilecki r48382 trunk/target/linux/ bcm53xx/patches-4.4/800-Revert-PCI-iproc-Add-PAXC-interface-support.patch bcm53xx/patches-4.1/800-Revert-PCI-iproc-Add-PAXC-interface-support.patch * bcm53xx: revert iProc patch adding PAXC support Jan 19 22:55:08 mamarley: i found sth regarding pty - http://wiki.beyondlogic.org/index.php?title=Cross_Compiling_BusyBox_for_ARM mentions some glibc kernel dropbear issue with pty on some configs Jan 19 22:57:17 at least on x86 i don't see any dropbear login issues with mamarley's patch Jan 19 22:57:39 I tested on x86_64 and didn't get any issues there either. Jan 19 22:59:10 swalker: plntyk: On which platforms were you guys testing? Jan 19 22:59:32 nbd r48383 trunk/ (7 files in 4 dirs) * toolchain: Add glibc 2.22 Jan 19 23:01:11 mamarley: i just booted with x86_64 and qemu and it was *almost* fine (i forgot some depends probably and got an libjson-c missing error) Jan 19 23:02:49 I was testing on a fresh, almost-default configuration for x86_64. The only changes I made were to build a VDI image and to use binutils 2.25, libstdc++ instead of uclibc++, and glibc 2.22. Jan 19 23:09:04 mine was octeon, I won't care when musl gains mips64 support, doing a fresh build Jan 19 23:10:11 got other changes in my tree, uhttpd/ustream-ssl/polarssl is generally borked atm Jan 19 23:10:49 swalker: details? Jan 19 23:16:58 nbd: http://pastebin.com/raw/trStnq27, CNs are completely bogus, uhttpd px5g config is correct Jan 19 23:19:10 huh, weird Jan 19 23:19:35 ECDHE-RSA-AES256-GCM-SHA384 - it's not supposed to even use that Jan 19 23:22:22 nbd: I just tested the restore-mark mask of 0x0f that you recommended and it does indeed work correctly that way. I will submit a patch. Jan 19 23:25:29 swalker: i can't reproduce that Jan 19 23:26:18 jow r48384 trunk/package/network/ipv6/ 6in4/Makefile 6in4/files/6in4.sh * Revert "6in4: Corrected tunnelbroker tunnel update URL" Jan 19 23:27:55 nbd r48385 trunk/package/network/services/uhttpd/files/uhttpd.config * uhttpd: fix typo in default config for px5g Jan 19 23:29:46 nbd: that's right, hence the other changes, http://pastebin.com/raw/qcpnLu5G Jan 19 23:31:03 so the issue only happens with those changes? Jan 19 23:31:45 need to investigate more Jan 19 23:32:55 swalker: standalone px5g or dynamically linked one? Jan 19 23:33:27 jow_laptop: dynamic Jan 19 23:33:36 jow_laptop: i have a commit for replacing busybox wget with uclient-fetch. it works well with opkg in my test, and if i install ca-certificates and ustream-polarssl, opkg can download from https://downloads.openwrt.org as well Jan 19 23:33:55 should i push it, or do you want to test first? Jan 19 23:34:42 nbd: push it. I'll likely not be able to test it soon Jan 19 23:34:48 ok Jan 19 23:35:45 nbd r48386 trunk/package/ system/opkg/Makefile utils/busybox/Config-defaults.in * opkg: depend on uclient-fetch, disable busybox wget Jan 19 23:41:48 does "wget" symlnk to uclient-fetch with some similar arguments? (just for my own script curiosity) Jan 19 23:42:26 uclient-fetch is command line compatible to wget Jan 19 23:42:29 nbd: debug check a few line below in r48363 is cruft now Jan 19 23:43:29 debug check? Jan 19 23:43:45 nbd: https://dev.openwrt.org/browser/trunk/toolchain/musl/common.mk?rev=48363#L43 Jan 19 23:43:56 oh, right Jan 19 23:44:35 command line compatible being, callable as "wget" or just the same arguments? Jan 19 23:45:24 nbd r48387 trunk/toolchain/ musl/Config.in Config.in musl/common.mk * toolchain/musl: remove old CONFIG_MUSL_ENABLE_DEBUG option Jan 19 23:45:29 thanks Jan 19 23:45:29 ^ https://dev.openwrt.org/changeset/48379 Jan 19 23:45:39 karlp: callable as wget, accepts the same command line arguments Jan 19 23:46:17 i even added the same progress bar code ;) Jan 19 23:47:09 there's only one thing that i didn't implement Jan 19 23:47:11 ftp support Jan 19 23:47:27 i don't expect many people are using that Jan 19 23:47:30 and ftp is a horrible protocol anyway Jan 19 23:48:11 busybox wget does not support ftp either Jan 19 23:48:11 https://dev.openwrt.org/browser/trunk/toolchain/glibc/common.mk?rev=48383#L76 still wanted for glibc 2.22? Jan 19 23:49:18 jow_laptop: from the code it looks like it does Jan 19 23:49:31 haven't tested if it actually works though Jan 19 23:50:15 swalker: Oops, how did I miss that? I grepped the entire source tree for references to VERSION_2_21. I will fix it, just a sec. Jan 19 23:54:23 Patch sent, sorry. Jan 19 23:55:24 i think it would probably be better to reverse this condition Jan 19 23:55:32 and check for eglibc Jan 19 23:55:49 we should probably also do this for the other checks Jan 19 23:55:59 to avoid running into the same crap again with the next release Jan 19 23:56:18 OK, one second. Jan 19 23:56:57 I can actually reverse all of those conditionals. Jan 19 23:57:15 nbd r48388 trunk/package/network/config/qos-scripts/files/usr/lib/qos/generate.sh * qos-scripts: Fix classification of ingress traffic Jan 20 00:05:11 nbd: I reversed two of them but I can't see any good way to reverse the ones starting on line 369 of package/libs/toolchain/Makefile. Is that OK? Jan 20 00:08:21 maybe $(if $(if $(CONFIG_EGLIBC_USE_VERSION_2_19),$(CONFIG_EGLIBC_OPTION_EGLIBC_UTMP),y),libutil) Jan 20 00:09:23 Yeah, that should work. Sorry, my GNU Make is a bit rusty. Jan 20 00:09:36 nor Jan 20 00:09:38 no problem Jan 20 00:14:48 nbd: Wait, wouldn't that cause problems for musl and uclibc? Previously, if musl or libc was selected, the if would evaluate to false, causing libutil not to be added. Now, it will evaluate to true. Wouldn't that cause problems? Jan 20 00:18:25 mamarley: musl and uclibc don't use that Jan 20 00:18:44 i'll get some sleep now. bbl. Jan 20 00:18:48 So it will be set, but will have no effect? OK. Jan 20 00:20:38 nbd: thanks, just as long as the basic "wget" still existed, couldn't care less about the progress bar myself. Jan 20 00:20:49 what's the advantages of uclient-fetch if I may ask? Jan 20 00:25:19 nbd good evening, can I ask some kernel questions in pm? I don't mind asking here, but I think it's rather offtopic hehe Jan 20 00:28:19 heh, no wonder I couldn't find uclient, I wasn't looking in "libs" :) Jan 20 01:44:36 any news on that hfsc stuff? **** ENDING LOGGING AT Wed Jan 20 02:59:59 2016