**** BEGIN LOGGING AT Fri Jan 31 02:59:58 2014
Jan 31 07:29:28 <Hopscotch> good morning
Jan 31 07:33:10 <builder_> build #510 of orion is complete: Failure [failed compile_8]  Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/510
Jan 31 10:46:16 <OpenWrt-Commits> nbd r39431 trunk/package/network/services/hostapd/files/netifd.sh * hostapd: fix basic rate list handling with netifd
Jan 31 11:08:19 <OpenWrt-Commits> nbd r39432 trunk/package/kernel/mac80211/patches/513-ath9k_add_pci_ids.patch * ath9k: fix handling of the default chip pci id on ar93xx (#14886)
Jan 31 11:11:38 * russell-- seeing a couple of these on boot of an ubnt airrouter (ar71xx): BUG: sleeping function called from invalid context at kernel/mutex.c:95
Jan 31 11:14:24 <russell--> http://sprunge.us/AAOg
Jan 31 11:14:33 <russell--> something to do with swconfig?
Jan 31 11:19:57 * russell-- had turned on that debugging feature via kernel_menuconfig, fwiw
Jan 31 11:25:59 <OpenWrt-Commits> cyrus r39433 trunk/package/network/ipv6/odhcp6c/Makefile * odhcp6c: several bugfixes and improvements
Jan 31 11:31:57 <karlp> thinks like CONFIG_BUSYBOX_CONFIG_DIFF=y that I used in Attitude Adjustment, and ws under the menuconfig of Base System->Busybox, where is that now in trunk?
Jan 31 11:34:49 <plntyk> it was removed
Jan 31 11:35:40 <plntyk> 39107 and other commits - to change busybox config you have to use the busybox config and copy it to env/ or so
Jan 31 11:35:57 <plntyk> there are posts about this on the mailing list and in the forum
Jan 31 11:36:24 <karlp> hrm, I don't remember seeing this on the mailling list, must have been a busy day for me :|
Jan 31 11:36:31 <karlp> ok,I'll go do some reading
Jan 31 11:36:49 <plntyk> minor shitstorm in the forum like "who would do such a thing"
Jan 31 11:36:53 <plntyk> :)
Jan 31 11:37:15 <karlp> I'm going to refrain from statements until I've read a bit more :)
Jan 31 11:38:24 <karlp> if nbd wants to avoid bogus bug reports from bad cases of user tinkering, he should prevent people from checking out the source and building it themselves :)
Jan 31 11:43:58 <plntyk> yes - but its like several other stuff in the buildroot - you have alway check if it does not occur with single job build after a distclean
Jan 31 11:45:24 <plntyk> and like 3 occasions where i found differences in packages depending on host linux
Jan 31 11:45:47 <karlp> so, if busybox config was removed, why is there still the NFS mount support option?
Jan 31 11:46:11 <plntyk> i think this option is broken
Jan 31 11:47:02 <plntyk> https://dev.openwrt.org/ticket/14636 - some strange issue there
Jan 31 11:47:34 <plntyk> silent cross compile fails :)
Jan 31 12:07:30 <karlp> ok, I think I've read enough.  I'm not personalyl _for_ removing this option, but if it's been removed, whoever removed it should please update http://wiki.openwrt.org/doc/howto/build to include the recommended steps for doing this sort of thing now
Jan 31 12:15:55 <russell--> karlp: CONFIG_PACKAGE_diffutils=y ?
Jan 31 12:17:01 <karlp> that's a substantially larger package, but yes, it would provide me with diff.
Jan 31 12:17:56 <karlp> plntyk: yeah, I added two config options into env/busyboxconfig and I get the same missing rpc/rpc.h
Jan 31 12:20:45 <plntyk> yeah - i've been using a github mirror of openwrt and of packages to manage all small improvements and bug fixes
Jan 31 12:21:10 <plntyk> there are like 50(?) packages where the download link is broken
Jan 31 12:21:33 <karlp> rpc/rpc.h missing doesn't sound like a download link problem...
Jan 31 12:23:53 <jow_laptop> karlp: Base System -> busybox -> Enable NFS mount support
Jan 31 12:24:38 <karlp> menuconfig: off,  with it =y in the env/busybox-config didn't build, menuconfig: on with =y in the env/busybox-config didn't build either way for me
Jan 31 12:25:03 <jow_laptop> if its enabled in menuconfig it will set CONFIG_BUSYBOX_ENABLE_NFS_MOUNT
Jan 31 12:25:11 <jow_laptop> which in turn should cause librpc to get enabled
Jan 31 12:25:31 <jow_laptop> (Selected by: PACKAGE_busybox [=y] && BUSYBOX_ENABLE_NFS_MOUNT)
Jan 31 12:26:33 <karlp> well it doesn't seem to.  diffconfig after enabling it in make menuconfig doesn't show anything about rpc
Jan 31 12:27:56 <jow_laptop> yeah, dep is broken
Jan 31 12:28:10 <karlp> what on earth is this? http://fpaste.org/73360/13911712/
Jan 31 12:28:48 <jow_laptop> try this please: http://pastebin.com/rtqX59TV
Jan 31 12:29:17 <jow_laptop> (followd by defconfig)
Jan 31 12:29:48 <karlp> shows up in diffconfig now at least :)
Jan 31 12:30:03 <jow_laptop> recompiling busybox should now trigger compilation of librpc
Jan 31 12:30:14 <jow_laptop> which in turn should prvde the previously missing rpc.h header
Jan 31 12:30:32 <karlp> yeah, it's building now, showing up in diffconfig is good enough proof for me.
Jan 31 12:30:40 <karlp> yup, built fine now.
Jan 31 12:31:27 <OpenWrt-Commits> jow r39434 trunk/package/utils/busybox/Makefile * busybox: fix dependency on librpc (#14636)
Jan 31 12:32:50 <karlp> that ticket can be closed now then I'd say.
Jan 31 12:35:12 <karlp> I'd really like to see some way of graphically editing the busybox settings again, make -C build_dir/target-*/busybox* menuconfig is functional, but not real obvious
Jan 31 12:35:28 <jow_laptop> I hate the change as well
Jan 31 12:39:28 <karlp> hwo fixed is stone is it? is it worth trying to dcoument the state as it is now? or does that just constitute tacit approval?
Jan 31 12:40:37 <jow_laptop> I can somewhat understand the reasoning, but in the end it might have been more worthwhile to add busybox applet depends to our stuff
Jan 31 12:40:49 <jow_laptop> instead of preventing bb configuration
Jan 31 12:41:31 <jow_laptop> that would have avoided cases like the missing readlink support in old configs that prevented wifi config on recent trunk versions
Jan 31 12:41:57 <nbd> i'm open to reverting the change, given the fallout
Jan 31 12:42:11 <nbd> if you guys prefer this as a solution
Jan 31 12:42:34 <jow_laptop> I could lend a hand with the applet depends
Jan 31 12:42:49 <nbd> i think making the change was useful in that it revealed how many users are doing much weird stuff with the busybox config
Jan 31 12:43:10 <jow_laptop> KanjiMonster suggestion of doing some taining sounded good as well
Jan 31 12:43:39 <jow_laptop> essentially hide the bb config behind a symbol that taints the version number enabled
Jan 31 12:43:45 <jow_laptop> *when enabled
Jan 31 12:43:49 <nbd> it would be nice if we could find a way to enable busybox customization through a separate option
Jan 31 12:43:53 <nbd> and make that enable the taint
Jan 31 12:44:11 <nbd> and without that option enabled, force busybox to defaults
Jan 31 12:45:20 <nbd> one way to do this would be to change the generated menuconfig code to add defaults similar to package preselection
Jan 31 12:45:24 <nbd> with separate DEFAULT_* symbols
Jan 31 12:46:01 <nbd> and in the case of no busybox customization enabled, generate the config entirely from the DEFAULT_* symbols
Jan 31 12:46:04 <jow_laptop> what is speaking against actually depending on applets from base-files and package/mac80211 ?
Jan 31 12:46:26 <jow_laptop> that should prvent users from deselecting them
Jan 31 12:46:51 <nbd> i guess we could add some of that, but i'm not really interested in working on maintaining a comprehensive list there
Jan 31 12:46:52 <jow_laptop> should be not more tedious than maintaining the defconfig
Jan 31 12:47:19 <nbd> busybox is the kind of thing that you should usually only customize if you have some idea about what you're doing
Jan 31 12:47:48 <nbd> and i want to make sure that people that don't customize busybox always get the up-to-date defaults
Jan 31 12:49:21 <nbd> i think we could probably write a script that turns a busybox .config into a Config.in files containing config BUSYBOX_DEFAULT_<symbol> \n default <value>
Jan 31 12:49:54 <nbd> and unconditionally add 'default BUSYBOX_DEFAULT_<symbol>' to all selectable config symbols in the translated busybox Config.in files
Jan 31 12:50:03 <nbd> that way we have clean separation between defaults and the menuconfig code
Jan 31 12:50:11 <nbd> and can hide the entire menu behind a single 'if' line
Jan 31 12:50:19 <nbd> what do you think?
Jan 31 12:50:45 <nbd> it has the added advantage of making it easier to update the defaults with busybox version updates, because you can just run menuconfig in the busybox dir and update the defauls file via script
Jan 31 12:50:56 <jow_laptop> yes, that'd be a solution
Jan 31 12:51:24 <jow_laptop> we can then incorperate the busybox config symbol in include/version.mk to taint the release numbers
Jan 31 12:52:04 <nbd> we still have to figure out how to handle package dependencies on busybox symbols then
Jan 31 12:52:14 <nbd> because they need to select the default and optionally the symbol itself
Jan 31 12:52:30 <nbd> could just add a wrapper macro for that
Jan 31 12:53:32 <nbd> e.g. DEPENDS += $(call busybox_config,FEATURE_DF_FANCY UNCOMPRESS)
Jan 31 12:53:47 <jow_laptop> hm
Jan 31 12:54:10 <jow_laptop> can't we extend whatever handles @sym to do treat @BUSYBOX_ special?
Jan 31 12:54:40 <nbd> i guess that would make backwards compat easier, but would be more hackish in the implementation
Jan 31 12:54:55 <nbd> and it would be confusing wrt. conditional deps
Jan 31 12:55:03 <nbd> since conditional deps are needed internally
Jan 31 12:55:25 <nbd> or another way would be to select only the default symbols
Jan 31 12:55:42 <nbd> and people that customize busybox need to ensure that the config has the necessary stuff enabled
Jan 31 12:55:59 <jow_laptop> yes
Jan 31 12:56:24 <nbd> it only matters when selecting a package that selects busybox symbols, after the busybox customization was already enabled
Jan 31 12:56:37 <nbd> if busybox customization is enabled after the extra package, it will carry the right defaults
Jan 31 12:56:49 <jow_laptop> imo packages shouldn't select bb features anyway
Jan 31 12:57:02 <nbd> i agree
Jan 31 12:57:03 <jow_laptop> side effects that result in conditional compilation are strongly discouraged
Jan 31 12:57:20 <nbd> just mentioning it, because people use custom packages to customize their busybox config
Jan 31 12:57:30 <jow_laptop> right
Jan 31 12:58:03 <jow_laptop> I guess we could also override Package/busybox/configure and emit a few warnings if stuff is not selected that should be default
Jan 31 13:30:36 <nbd> jow_laptop: i already adjusted the menuconfig generator code and added support for the defaults stuff
Jan 31 13:30:50 <nbd> now i just need to hook it into the code that generates busybox .config
Jan 31 13:47:19 <nbd> jow_laptop: i will commit my reworked busybox menuconfig implementation, will you take care of the taint stuff?
Jan 31 13:49:28 <jow_laptop> yes, can look into it
Jan 31 13:53:17 <OpenWrt-Commits> nbd r39435 trunk/package/ (54 files in 26 dirs)
Jan 31 13:53:18 <OpenWrt-Commits> busybox: add a reworked implementation of menuconfig support, this time with a guard option that keeps all symbols at default values until an extra option is activated
Jan 31 13:55:40 <jow_laptop> how do we want to present the taints?
Jan 31 13:55:49 <jow_laptop> "+bb" in the version tag?
Jan 31 13:56:21 <jow_laptop> e.g. DISTRIB_REVISION="r39435+bb"
Jan 31 13:56:41 <KanjiMonster> I'd put a space between them
Jan 31 13:57:06 <jow_laptop> r39435 +bb ?
Jan 31 13:57:49 <jow_laptop> also do we want to amend the busybox package version?
Jan 31 13:58:20 <jow_laptop> e.g. PKG_RELEASE += $(if $(CONFIG_BUSYBOX_CUSTOM),+custom,)
Jan 31 13:58:42 <KanjiMonster> is it that common that people use a different version?
Jan 31 13:59:10 <jow_laptop> its about making it easily spottable that bb was modified
Jan 31 13:59:32 <jow_laptop> right now bb from the snasphot or release builds and bb with some custom changes share exactly the same package version
Jan 31 13:59:43 <KanjiMonster> right
Jan 31 14:00:08 <KanjiMonster> but the bb-version can be found out by opkg list_installed; the used busybox config not
Jan 31 14:00:35 <jow_laptop> so when someone modifies bb, the version string would be "1.19.4-7+custom"
Jan 31 14:00:48 <jow_laptop> as opposed to "1.19.4-7"
Jan 31 14:01:29 <jow_laptop> this is mainly for bug reports
Jan 31 14:01:42 <jow_laptop> when you ask people to provide a list of their installed packages etc.
Jan 31 14:01:54 <jow_laptop> /etc/openwrt_{release,version} might be customized through menuconfig
Jan 31 14:02:50 <KanjiMonster> maybe instead of appending them to the version add an extra field DISTRIB_TAINTS or so that can't be customized?
Jan 31 14:06:55 <jow_laptop> http://pastebin.com/bcz6Qn4g
Jan 31 14:09:49 <KanjiMonster> let me put together what I had when I poked at it
Jan 31 14:11:10 <danitool> KanjiMonster: I've connected a SPI flash memory to my old livebox (bcm6348)
Jan 31 14:11:44 <danitool> it's detected OK, but when reading or writing i get this message
Jan 31 14:11:47 <danitool> m25p80 spi0.0: unable to do transfers larger than FIFO size (260 > 63)
Jan 31 14:13:15 <danitool> can this be solved with some work at the bcm63xx driver?
Jan 31 14:13:55 <KanjiMonster> jow_laptop: my hackish WIP is this -> http://pastebin.com/c3kxFCbH
Jan 31 14:14:43 <KanjiMonster> (funny that we both use %t)
Jan 31 14:16:25 <jow_laptop> hm, do you consider CONFIG_ALL to be a taint?
Jan 31 14:16:36 <jow_laptop> and can you explain the menu.c change?
Jan 31 14:16:57 <KanjiMonster> that was working around a segfault for hidden menus
Jan 31 14:17:15 <KanjiMonster> it's directly pulled from upstream kernel
Jan 31 14:17:19 <jow_laptop> ah ok
Jan 31 14:17:52 <KanjiMonster> config allows adding a "visible if <bar>" condition to menus
Jan 31 14:18:29 <KanjiMonster> but it didn't copy that do submenus, which then caused a segfault for nested hidden menus
Jan 31 14:19:22 <KanjiMonster> I needed it because I played around how I can easily make the visibilty of BB options conditional without removing the config options from .config itself
Jan 31 14:20:30 <KanjiMonster> jow_laptop: probably rather being ALL not set should be the taint ("no-all"), because that is usually corelated to dysfunctional kernel modules
Jan 31 14:21:50 <jow_laptop> yeah, idea of a better name for "no-all" ?
Jan 31 14:22:57 <KanjiMonster> unfortunately not
Jan 31 14:24:01 <KanjiMonster> there are several things that might make sense as no-foo or positive taings, e.g. ipv6 support
Jan 31 14:24:08 <KanjiMonster> *taints
Jan 31 14:24:17 <jow_laptop> what about "-all" and "-ipv6" ?
Jan 31 14:24:31 <jow_laptop> or "!all" and "!ipv6"
Jan 31 14:24:53 <KanjiMonster> !i looks awkward
Jan 31 14:25:10 <jow_laptop> and for the others we can use +
Jan 31 14:25:16 <jow_laptop> "+bb", "+mklibs"
Jan 31 14:26:18 <jow_laptop> so it'd look like BARRIER BREAKER (Bleeding Edge, r38991 -all -ipv6 +mklibs +bb)
Jan 31 14:26:55 <KanjiMonster> my biggest gripe with that is that my inner OCD then wants to include them always, either with + or - prefixed depending on their selection state ;)
Jan 31 14:27:46 <KanjiMonster> I'm not yet sure whether all makes sense as it only changes defaults, but does not necessarily change actual options
Jan 31 14:28:16 <KanjiMonster> e.g. if you select all with a preexisting config it will then pickup +all despite the build not being +all at all
Jan 31 14:29:00 <jow_laptop> I'd rather leave out CONFIG_ALL for now
Jan 31 14:29:05 <KanjiMonster> yeah
Jan 31 14:29:18 <jow_laptop> just ipv6, mklibs, busybox and maybe eglibc
Jan 31 14:29:29 <KanjiMonster> right
Jan 31 14:31:00 <KanjiMonster> using svn/git we could try to find out if there are modified files/additional commits, maybe that could trigger a dirty flag or so
Jan 31 14:31:15 <jow_laptop> that sounds like it belongs into getver.sh
Jan 31 14:31:29 <KanjiMonster> yeah
Jan 31 14:35:53 <KanjiMonster> jow_laptop: this is what I did to busybox that needed the kconfig patch (but it didn't work as expected) => http://pastebin.com/g7KBb46u
Jan 31 14:36:07 <KanjiMonster> (based on an older busybox revision that still had its menu)
Jan 31 15:11:22 <danitool> KanjiMonster: any idea about my question?
Jan 31 15:13:36 <KanjiMonster> danitool: you need to extend the m25p80 limited buffer hack to also apply to writes and pass it through the flash platform_data
Jan 31 15:24:05 <karlp> jow_laptop: KanjiMonster's tainting the version sounds like a nice solution, that would then also apply to other "fun" things like changing libc versions and so on right?  there's more than just busybox in ways to break things :)
Jan 31 15:28:51 <karlp> nvm, did more backlog reading, sounds like yo've got this handled.
Jan 31 15:29:16 <karlp> an update to the wiki page on building openwrt and/or the creating packages would be nice for some of this when it's final
Jan 31 15:51:42 <OpenWrt-Commits> jow r39436 trunk/include/version.mk
Jan 31 15:51:42 <OpenWrt-Commits> version.mk: add initial infrastructure for recording specific build taint conditions like modified busybox or disabled ipv6 support
Jan 31 15:51:46 <OpenWrt-Commits> jow r39437 trunk/package/base-files/ Makefile files/etc/openwrt_release * base-files: expose taint flags in /etc/openwrt_release
Jan 31 15:56:51 <OpenWrt-Commits> wigyori r39438 trunk/ package/boot/uboot-sunxi/uEnv.txt package/boot/uboot-sunxi/Makefile target/linux/sunxi/config-3.12
Jan 31 15:56:52 <OpenWrt-Commits> sunxi: boot changes
Jan 31 15:56:52 <OpenWrt-Commits>  - Added uEnv.txt to facilitate automatic boot.
Jan 31 15:56:52 <OpenWrt-Commits>  - Cosmetic fix in u-boot Makefile
Jan 31 15:56:52 <OpenWrt-Commits>  - Don't force command line arguments
Jan 31 15:56:55 <OpenWrt-Commits> Signed-off-by: Zalan Blenessy <zalan.blenessy@gmail.com>
Jan 31 15:56:57 <OpenWrt-Commits> Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
Jan 31 16:07:18 <OpenWrt-Commits> wigyori r39439 trunk/target/linux/generic/config-3.13 * [3.13]: add missing symbol
Jan 31 17:25:36 <zajec_> could someone explain me, how multiple wireless interfaces (virtual interfaces) work on the low level?
Jan 31 17:26:02 <zajec_> does hardware sends packets with different VLAN tags to the system?
Jan 31 17:26:22 <zajec_> or are packet VLAN-tag-less and they are splitted by system using something like SSID in the packet?
Jan 31 17:26:55 <KanjiMonster> zajec_: I would assume #linux-wireless is a better place for this question ;)
Jan 31 17:27:15 <zajec_> hm, maybe :)
Jan 31 17:28:34 <KanjiMonster> of course this one isn't a bad place, but the other one might be better ;)
Jan 31 17:29:23 <zajec_> i found following article: http://www.juniper.net/techpubs/software/junos-security/junos-security10.0/junos-security-swconfig-wlan/wlan-ax411-virtual-access-point-vlan-understanding.html
Jan 31 17:29:32 <zajec_> "When a wireless client connects to the access point, the access point tags traffic from the client with a VLAN ID."
Jan 31 17:29:50 <zajec_> but it sounds unlikely... every client gets a separated VLAN ID? sounds insane
Jan 31 17:30:13 <zajec_> ah, ok
Jan 31 17:30:17 <zajec_> it's not one VLAN per client
Jan 31 17:30:30 <plntyk> one vlan per AP is some policy or so
Jan 31 17:31:44 <zajec_> so... if virtual WLANs are actually different VLANs, why OpenWrt usues wlan0-0, wlan0-1, wlan-2, etc.? instead of wlan0.1, wlan0.2, wlan0.3?
Jan 31 17:32:57 <zajec_> "The same VLAN can be configured for multiple virtual access points." ok, maybe that's why
Jan 31 17:35:44 <plntyk> all those different wlan features seem strange - it feels like nothing is really working because everybody and his brother is working on some new feature , trying to get some new standard established or so
Jan 31 17:36:29 <plntyk> like those different P2P modes ... didnt have time to look at them
Jan 31 17:36:46 <plntyk> and P2P sounds like some "adhoc" reinvented
Jan 31 17:37:59 <plntyk> and those closed firmware makes the hole thing more crazy because now some firmware might have some WLAN feature disabled
Jan 31 17:38:23 <plntyk> afaik AP only firmware of ath10k has no adhoc
Jan 31 17:40:48 <plntyk> oh .. a new 802.11ac patch to play with on the weekend :)
Jan 31 17:59:00 <karlp> new busybox looks pretty nice guys :)
Jan 31 17:59:11 <karlp> here's some help text that might go with it: http://paste.fedoraproject.org/73456/91191113/
Jan 31 18:00:14 <karlp> damnit, typo: updated: http://paste.fedoraproject.org/73457/19119913
Jan 31 18:03:36 <karlp> ps
Jan 31 19:13:27 <jow_laptop> karlp: can you make a proper patch with Sob?
Jan 31 19:13:34 <jow_laptop> I'll merge it then
Jan 31 19:16:56 <karlp> sorry, my bad: http://paste.fedoraproject.org/73470/39119580/
Jan 31 19:18:13 <OpenWrt-Commits> jow r39440 trunk/package/utils/busybox/Config.in * busybox: Add help documentation in menuconfig
Jan 31 23:24:53 <builder_> build #472 of rb532 is complete: Failure [failed shell_12]  Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/472
Jan 31 23:25:54 <builder_> build #472 of ppc44x is complete: Failure [failed shell_12]  Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/472
**** ENDING LOGGING AT Sat Feb 01 02:59:58 2014