**** BEGIN LOGGING AT Mon Oct 05 02:59:58 2015 Oct 05 06:52:15 build #105 of cobalt is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/105 Oct 05 06:54:57 build #105 of cns21xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cns21xx/builds/105 Oct 05 06:57:43 build #105 of orion is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/105 Oct 05 08:29:01 can someone have a look on this and provide feedback on the design? https://github.com/txomon/libubox Oct 05 08:30:40 binary.h Oct 05 08:32:19 i'm not entirely convinced that it's useful Oct 05 08:32:29 also, it's hard to understand how the functions work because there are no examples Oct 05 08:33:07 it seems to me that a lot of the functionality should be handled by properly written structs instead Oct 05 08:45:11 neither I am convinced about the usability. but it has some good features towards writing/reading endianness Oct 05 08:45:18 let me complete with some examples Oct 05 08:55:54 rmilecki r47107 trunk/target/linux/ar71xx/image/Makefile * ar71xx: rename firmware for Chinese version of TL-WR941ND v6 Oct 05 09:11:43 jow r47108 trunk/package/network/utils/iwinfo/Makefile * iwinfo: nl80211: add support for reading TX power from netlink Oct 05 09:21:22 nbd: https://github.com/txomon/libubox/blob/master/binary.h#L72-L96 Oct 05 09:26:32 txomon: ok, i still don't understand the API then Oct 05 09:26:47 especially the 'read' part Oct 05 09:26:56 what's the connection between the struct, the format and the raw data Oct 05 09:27:43 the struct is supposed to be the one that would contain that raw formatted data Oct 05 09:28:24 then why are you doing my_data = readbv(v, "%lw%2i%ay%p4d"); with my_data being a struct? Oct 05 09:28:32 1 uint16_t, 2 bits, 6 unused bits, 1 chat/byte, 4 32bit words Oct 05 09:28:59 I was trying to show how to load to an struct Oct 05 09:29:06 from what? Oct 05 09:29:15 from the memory blobl Oct 05 09:29:23 there is endianess to be taken care of Oct 05 09:29:23 what kind of blob? Oct 05 09:29:30 the one created on writeb() Oct 05 09:29:37 so the format of the blob is the same as the struct, except that readbv does an endian swap? Oct 05 09:29:47 and the write too Oct 05 09:29:58 write makes endian conversion too Oct 05 09:30:13 but yes, the blob is the same as the struct Oct 05 09:30:39 the thing about readb and readbv is that it allows you to load blobs to different memory pointers Oct 05 09:30:51 or to the same, but taking care of alignment Oct 05 09:31:13 anyway I am not sure if it's worth the effort Oct 05 09:31:30 i think this is inferior to just dealing with the struct directly Oct 05 09:31:44 the big problem is that you suddenly have two vastly different representations of the same data format Oct 05 09:31:57 and nothing is checking whether they are in sync Oct 05 09:32:07 making this whole thing very fragile Oct 05 09:32:10 true Oct 05 09:32:57 well, I was thinking on creating an enum and an const char * to hold all the format strings Oct 05 09:32:58 a better approach would be to make libubox accept kernel-style endian annotated data types for structs Oct 05 09:33:06 and add support for using sparse to validate endian access Oct 05 09:33:37 I like that Oct 05 09:33:48 will browse and come back to see how it does that Oct 05 09:34:10 although I am afraid I don't have enough knowledge Oct 05 09:34:36 it's just a bunch of #defines and some sparse rules to check them Oct 05 09:36:36 ok Oct 05 09:36:46 I am reading about them at https://lwn.net/Articles/205624/ Oct 05 10:04:09 hello people Oct 05 10:09:44 nbd: the broadcast thing seems like wont be accepted because the rest of the code doesn't use it :/ Oct 05 10:10:01 ? Oct 05 10:10:08 talking to john Oct 05 10:11:03 which makes sense,but anyway a pity for me Oct 05 10:12:04 for you it's not a big problem anyway, you can just put the setsockopt call in your application call after calling usock() Oct 05 10:26:31 yep Oct 05 10:26:55 blogic r47109 trunk/target/linux/mpc85xx/Makefile Oct 05 10:26:55 target/mpc85xx: check if 'generic' subtarget for image file cuImage.tl-wdr4900-v1 Oct 05 10:27:07 blogic r47110 trunk/target/linux/ar71xx/base-files/lib/ar71xx.sh * ar71xx: Correct WeIO board name Oct 05 10:27:14 blogic r47111 trunk/target/linux/ (5 files in 4 dirs) * ramips: Add base-files for HiWiFi HC5x61 models Oct 05 10:27:20 blogic r47112 trunk/target/linux/ ramips/dts/HC5XXX.dtsi ramips/dts/HC5661.dts ramips/dts/HC5861.dts ramips/dts/HC5761.dts * ramips: Add DTS files for HiWiFi HC5x61 models Oct 05 10:27:29 blogic r47113 trunk/target/linux/ ramips/image/Makefile ramips/mt7620/profiles/hiwifi.mk * ramips: Add build files for HiWiFi HC5x61 models Oct 05 10:27:40 blogic r47114 trunk/target/linux/ramips/files/drivers/net/ethernet/ralink/mt7530.c * ramips: Allow to receive vlan over untag ports on MT7530 Oct 05 10:27:48 blogic r47115 trunk/target/linux/ ramips/files/drivers/net/ethernet/ralink/ralink_soc_eth.h ramips/files/drivers/net/ethernet/ralink/ralink_soc_eth.c * ramips: Fix setting of rx buffer length Oct 05 10:27:55 blogic r47116 trunk/target/linux/ramips/files/drivers/net/ethernet/ralink/ralink_soc_eth.c * ramips: Fix too small rx buffer Oct 05 10:28:01 blogic r47117 trunk/target/linux/ramips/files/drivers/net/ethernet/ralink/gsw_mt7620a.c * ramips: Enable rx of 1536 bytes ethernet frames on MT7621 Oct 05 10:28:07 blogic r47118 trunk/target/linux/ramips/patches-3.18/0030-pinctrl-ralink-add-pinctrl-driver.patch * ramips: Ralink RT3883 RGMII pinmux fix. Oct 05 10:28:21 blogic r47119 trunk/target/linux/ (6 files) * ramips: fix size-cells for spi nodes Oct 05 10:28:26 blogic r47120 trunk/target/linux/ (5 files in 2 dirs) * ramips: add support for 16MB VoCore Oct 05 10:28:30 blogic r47121 trunk/target/linux/ramips/patches-3.18/0017-MIPS-use-set_mode-to-enable-disable-the-cevt-r4k-irq.patch * ramips: fix use remove_irq to release irqaction resource Oct 05 10:28:36 blogic r47122 trunk/target/linux/ramips/patches-3.18/0066-cevt.patch * ramips: improve systick timer Oct 05 10:28:48 blogic r47123 trunk/target/linux/brcm2708/modules.mk * brcm2708: fix kmod-sound-soc-bcm2708-i2s config Oct 05 10:29:02 blogic r47124 trunk/target/linux/ brcm2708/image/Makefile brcm2708/Makefile * brcm2708: switch device tree compilation to linux Oct 05 10:29:09 blogic r47125 trunk/target/linux/ brcm2708/image/Makefile brcm2708/image/config.txt * brcm2708: add support for device tree overlays Oct 05 10:29:16 blogic r47126 trunk/target/linux/brcm2708/base-files/ lib/upgrade/platform.sh lib/preinit/79_move_config * brcm2708: keep boot partition mounted to allow users to change config.txt Oct 05 10:29:24 blogic r47127 trunk/target/linux/lantiq/base-files/lib/functions/lantiq_dsl.sh * lantiq: bugfix uninitialized variables Oct 05 10:29:33 blogic r47128 trunk/target/linux/lantiq/ patches-4.1/0035-owrt-lantiq-wifi-and-ethernet-eeprom-handling.patch patches-4.1/0028-NET-lantiq-various-etop-fixes.patch * lantiq: Change the data-type of mac address in ETOP driver Oct 05 10:29:39 blogic r47129 trunk/target/linux/lantiq/ patches-4.1/0035-owrt-lantiq-wifi-and-ethernet-eeprom-handling.patch patches-3.18/0035-owrt-lantiq-wifi-and-ethernet-eeprom-handling.patch * lantiq: fix ath_pci_fixup hang on bridged devices Oct 05 10:29:46 blogic r47130 trunk/ (7 files in 5 dirs) * lantiq: Add target TP-Link TD-W8980 Oct 05 10:29:53 blogic r47131 trunk/target/linux/lantiq/Makefile * lantiq: make v4.1 the default kernel Oct 05 10:30:00 blogic r47132 trunk/target/linux/lantiq/ base-files/etc/uci-defaults/02_network base-files/etc/uci-defaults/01_leds dts/ARV7519RW22.dts * lantiq: several ARV7519RW22 fixes Oct 05 10:30:06 blogic r47133 trunk/package/boot/uboot-envtools/files/lantiq * uboot_envtools: add P2812HNUF1 Oct 05 10:30:13 blogic r47134 trunk/package/network/services/openvpn/files/openvpn.init * openvpn: add handling for route-pre-down option Oct 05 10:30:22 blogic r47135 trunk/ include/version.mk package/base-files/files/etc/device_info package/base-files/image-config.in * base-files: add URL option for OEM manufacturer info Oct 05 10:30:27 blogic r47136 trunk/package/network/services/lldpd/files/lldpd.init * lldpd: wrap procd command args in separate quotes Oct 05 10:38:56 build #104 of pxa is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/104 Oct 05 11:35:43 nbd r47137 trunk/target/linux/ramips/patches-3.18/0044-mtd-add-chunked-read-io-to-m25p80.patch * ramips: improve read performance of the m25p80 chunked io patch Oct 05 11:58:12 nbd: did you see my discussion with Ralf about setting ARCH_USE_BUILTIN_BSWAP for MIPS? Worth looking at on all platforms, I suspect. Oct 05 13:09:47 build #106 of brcm47xx.legacy is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx.legacy/builds/106 Oct 05 13:22:04 argh, tits. The 8139cp fixes are non-trivial to backport to 3.18 Oct 05 14:12:52 nbd r47138 trunk/include/prereq-build.mk * build: use host cflags/ldflags for openssl prereq check Oct 05 14:12:59 nbd r47139 trunk/include/host-build.mk * build: include homebrew include/library directory in cflags/ldflags on mac os x Oct 05 14:14:09 nbd r47140 branches/chaos_calmer/include/prereq-build.mk * build: use host cflags/ldflags for openssl prereq check Oct 05 14:14:12 nbd r47141 branches/chaos_calmer/include/host-build.mk * build: include homebrew include/library directory in cflags/ldflags on mac os x Oct 05 15:44:41 nbd r47142 trunk/package/kernel/mac80211/patches/338-mac80211-initialize-tid-field-in-struct-ieee80211_tx.patch * mac80211: add a fix to initialize the tid field in the tx queue struct Oct 05 16:26:27 rpcd is now failing to compile against glibc on both CC and trunk. Perhaps that and procd (which is also failing for the same reason) should be compiled with -Wno-error=unused-result? Oct 05 16:38:46 It also looks like the version number in the latest rpcd version is incorrect. It was changed from 2015-05-17 to 2015-01-10, which as far as opkg is concerned is a lower version number. Oct 05 16:42:04 nbd: broadcast doesn't work Oct 05 16:42:16 I am using usock and then setsockopt Oct 05 16:42:32 not working, I am afraid it needs to be done before bind() Oct 05 16:42:58 also, take into account that I am using the fd to be both server and client (dhcp like) Oct 05 16:49:36 txomon: what part is not working? Oct 05 16:49:40 does bind return an error? Oct 05 17:06:33 build #105 of ramips.mt7628 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7628/builds/105 Oct 05 19:13:31 build #105 of brcm47xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/105 Oct 05 20:29:23 build #105 of ramips.mt7621 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7621/builds/105 Oct 05 20:38:56 nbd: sending broadcast packets Oct 05 20:39:20 and I think bind works perfectly Oct 05 20:39:23 and you're sure that the setsockopt call worked? Oct 05 20:40:04 nbd: didn't check... but the code was basically main(){usock();setsockopt();sendto();} Oct 05 20:40:16 maybe I should rebind? Oct 05 20:45:03 i don't think you should rebind Oct 05 20:45:09 maybe throw the code in a pastebin somewhere Oct 05 21:03:24 ok, haven't got access now, but will do tomorrow Oct 05 21:03:32 although maybe I can replicate now... Oct 05 22:16:04 build #105 of ipq806x is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ipq806x/builds/105 Oct 05 22:31:33 nbd r47143 trunk/package/kernel/mt76/Makefile * mt76: update to the latest version, adds a few minor fixes Oct 05 22:39:59 nbd r47144 branches/chaos_calmer/package/kernel/mac80211/patches/398-mac80211-initialize-tid-field-in-struct-ieee80211_tx.patch * mac80211: add a fix to initialize the tid field in the tx queue struct Oct 05 22:40:06 nbd r47145 branches/ chaos_calmer/package/kernel/mt76/patches/100-dma-fix-rx-buffer-allocation-size.patch chaos_calmer/package/kernel/mt76/patches/002-disable_fast_xmit.patch chaos_calmer/package/kernel/mt76/patches/001-backport_ieee80211_hw_set.patch chaos_calmer/package/kernel/mt76/Makefile Oct 05 22:40:06 mt76: sync with trunk r47143, add backports for compat with the older mac80211 version Oct 06 00:56:08 is there a way to get the build to save all errors at the end instead of giving up on the first error? Oct 06 00:57:24 hrm, I guess I want "make -k".. anyone tried it? Oct 06 01:33:45 build #91 of mpc52xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/91 Oct 06 01:38:21 hm, I think "-k" isn't propagated to sub-makes, because my build still stopped on the first error **** ENDING LOGGING AT Tue Oct 06 02:59:58 2015