**** BEGIN LOGGING AT Tue Mar 23 02:59:57 2010 Mar 23 03:05:45 build #43 of brcm63xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/43 Mar 23 04:33:26 swalker * r20382 /packages/Xorg/lib/qt4/Makefile: [packages] qt4: add url Mar 23 04:35:06 swalker * r20383 /packages/libs/libowfat/Makefile: [packages] libowfat: add url, fix copyright, remove $Id$ Mar 23 04:42:10 ping crow Mar 23 05:17:57 build #29 of ar7 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar7/builds/29 **** BEGIN LOGGING AT Tue Mar 23 11:58:15 2010 Mar 23 11:58:19 the netfilter packages will enable them Mar 23 11:59:12 for synflood you need the ipt_limit module Mar 23 12:00:15 ... which should be part of the iptables core package Mar 23 12:00:33 is there a /usr/lib/xt_limit.so or ipt_limit.so on your device? Mar 23 12:01:05 /usr/lib/iptables/... I mean Mar 23 12:02:16 xMff: /usr/lib/iptables/libxt_limit.so is there Mar 23 12:02:21 the other one not Mar 23 12:02:57 this is okay Mar 23 12:03:10 and lsmod | grep xt_limit ? Mar 23 12:03:13 i think the main options for filtering are all set in the kernel configuration Mar 23 12:04:12 nope, xt_limit missing Mar 23 12:05:03 find out why Mar 23 12:05:08 I have to run now, sorry Mar 23 12:05:15 CONFIG_NETFILTER_XT_MATCH_LIMIT maybe its this one Mar 23 12:05:25 yes it is Mar 23 12:05:30 xMff: thank you anyway... :) Mar 23 12:05:37 xMff: happy running Mar 23 12:14:43 ping thepeople Mar 23 12:15:40 thepeople: your uci commit may have broken the the build process....I have reports of 404 on http://downloads.openwrt.org/sources/uci-12012009.5.tar.gz Mar 23 12:19:21 thepeople: build fails if the user has no git installed Mar 23 12:42:23 acoul * r20386 /trunk/target/linux/generic-2.6/config-2.6.31: generic-2.6: config-2.6.33 add missing symbols, closes #6913 (thanks maddes) Mar 23 13:22:33 cshore: ok, I will have to get it uploaded Mar 23 13:22:44 cshore: can you have them install git in the time being? Mar 23 13:22:52 thanks...they have Mar 23 14:32:18 nico * r20387 /trunk/package/linux-atm/Makefile: [package] linux-atm: fix compilation of 'qgen' host tool Mar 23 14:40:30 any plans for adding an x86-kvm target, and virtio support? or any objections against it? Mar 23 14:41:47 <{Nico}> stintel: waiting for a patch :) Mar 23 14:41:54 great, working on one ;-) Mar 23 14:42:56 I've noticed that xen network driver is put in the xen section rather than in the general network device section in menuconfig, so if I add a virtio category I better add the virtio-net driver there? coz right now I added it in netdevices.mk Mar 23 14:43:01 cshore: fixed Mar 23 14:43:18 thank you Mar 23 14:57:44 stintel: if you're going to create a kvm subtarget then you can merge xen.mk together with kvm stuff into a virtual.mk or something Mar 23 14:58:15 stintel: when I did the xen stuff it was the only paravirt target so I didn't bother to make it generic Mar 23 14:58:22 stintel: maybe add vmware in there as well Mar 23 15:02:31 acoul * r20388 /trunk/target/linux/generic-2.6/ (config-2.6.31 config-2.6.32 config-2.6.33): generic-2.6: fix r20386 (thanks nico), update missing symbols on config-2.6.32 too Mar 23 15:16:20 jow_laptop: I'll see what I can do Mar 23 15:16:48 thepeople: ^ Mar 23 15:16:48 :) Mar 23 15:17:36 :-) Mar 23 15:18:04 if someone could commit the oneline patch for xen.mk would be nice, then I dont have to send it again as part of patch series :) Mar 23 15:18:23 or w/e Mar 23 15:23:05 stintel: show me Mar 23 15:23:27 jow_laptop: I sent it to the ml only 10 minutes ago or so Mar 23 15:23:31 oh Mar 23 15:23:58 maybe I'd better just pastebin'd it :") Mar 23 15:24:58 jow * r20389 /trunk/package/kernel/modules/xen.mk: Mar 23 15:24:58 [PATCH] kernel: fix xen-fbdev description Mar 23 15:24:58 This patch fixes the description for KernelPackage xen-fbdev caused by a Mar 23 15:24:58 typo in xen.mk. Mar 23 15:24:58 Signed-off-by: Stijn Tintel Mar 23 15:25:31 thepeople: any update on 802.11w ? :) Mar 23 15:26:52 someone from cz offered to pay me to implement it in the ubiquiti 5.1 sdk, which looks a lot like 8.09.2, I told him he better used openwrt :-) Mar 23 15:27:35 and well "implement" is probably not the right word, it's already there, I only enabled it Mar 23 15:27:39 stintel: spaced it, to many other things to do, I will see if I can get to it tonight Mar 23 15:28:43 it might not apply cleanly anymore, haven't checked, but I have the git branch still around I could rebase it if needed (git is cool:) Mar 23 15:29:35 jow_laptop: the bot announces realtime or it was just coincidence ? Mar 23 15:30:08 stintel: kind-of-realtime Mar 23 15:30:26 sometimes it lags Mar 23 15:30:46 ok Mar 23 15:30:47 downhill wind behind it realtime Mar 23 15:30:49 :-P Mar 23 16:08:46 i've a ralink rt2561 wlan card. 'iwlist wlan0 scan' works, but the 'wifi detect' command has no output and the wireless options are not configurable in the luci webinterface Mar 23 16:10:22 qecko: you have to remove /etc/config/wireless for wifi detect to do something, afaik Mar 23 16:10:23 does someone know how to add support for this new card? /etc/config/wireless is empty. the wiki says that the 'type' is determined on firstboot, but where exactly? http://wiki.openwrt.org/doc/uci/wireless Mar 23 16:12:05 stintel: doesnt help, the DEVICES and DRIVERS parameters in the last lines of the /sbin/wifi script aren't defined too... Mar 23 16:15:10 jow_laptop: getting this with trunk r20389 / luci r5911: http://openwrt.pastebin.ca/1850482 , want me to file it @ luci trac ? Mar 23 16:18:17 hmm, seems some more stuff is broken, I'll make dirclean first Mar 23 16:18:54 stintel: hmm I fear that are the first fallouts from the uci update Mar 23 16:19:07 ah Mar 23 16:19:38 so this could be related as well: swconfig: can't load library 'libuci.so.0.7' Mar 23 16:20:04 yes Mar 23 16:20:59 stintel: http://luci.subsignal.org/trac/changeset/5655/luci/trunk/libs/lucid-http Mar 23 16:21:04 stintel: forgot to merge it Mar 23 16:21:44 I'll see if I can fix swconfig Mar 23 16:21:54 * thepeople grumbles, didn't test swconfig Mar 23 16:23:34 jow_laptop: thanks Mar 23 16:23:39 stintel: I merged the fix; ./sfripts/feeds update luci; make package/luci/clean ... Mar 23 16:28:16 thepeople: maybe the swconfig fails because I didn't make package/swconfig/clean Mar 23 16:28:52 I really need to get a quad core and always make clean first :) Mar 23 16:29:27 stintel: it seems likely Mar 23 16:29:41 stintel: quad core isn't fast enough Mar 23 16:29:43 :-P Mar 23 16:29:45 thepeople: :-) Mar 23 16:30:11 I only have an E8400 atm , want to replace it with Q9550 but no €€ atm :( Mar 23 16:31:31 stintel: yea, it isn't so bad when you aren't doing full builds Mar 23 16:35:34 xMff did you get that tcpdump? Mar 23 16:36:52 crow: yes but had no time to look into it yet Mar 23 16:37:31 jow_laptop ok thnx, well i asked in wrong channel maybe :) if more info is needed let me know, Mar 23 16:38:07 now I just need to figure out how to boot from flash instead usb stick (block-extroot) Mar 23 16:38:22 cshore1 maybe able to help you Mar 23 16:38:35 cshore1 ping Mar 23 16:38:41 yea i will ask him Mar 23 16:39:48 crow: hello Mar 23 16:40:11 cshore1 Hi, is ok to write here or i should in #openwrt Mar 23 16:42:08 here is good....you found a bug Mar 23 16:42:19 ok Mar 23 16:42:21 which one? Mar 23 16:44:04 anyway i tried block-extroot yesterday did used wrong ftsab at first place, but not i do reboot, wait until power led is of (wl500gpv1) then i plug out usb but it wont boot from flash (cant ssh,telnet to router), and every fev minutes its like router is in reboot cycle phase Mar 23 16:44:41 did i do something wrong or thats the scenarion, then the question is how to switch back to openwrt on flash Mar 23 16:45:02 crow: flash a new firmware Mar 23 16:45:30 while i am at usb stick openwrt? Mar 23 16:45:45 crow: oh, it boots with the stick? Mar 23 16:45:51 yes Mar 23 16:45:56 just without it no Mar 23 16:46:02 crow: you can sysupgrade -n Mar 23 16:46:21 crow: there is a bug that prevents the fallback to flash Mar 23 16:46:35 is that safe for wl500gpv1? "sysupgrade -n" Mar 23 16:46:36 crow: which I didn't see until you asked Mar 23 16:46:46 cshore1 :D so i found a bug :) Mar 23 16:47:18 crow: that will erase you current configuration, but should be safe, but I'm not an exper on wl500gpv1 Mar 23 16:47:50 cshore1 i used always this to flash new firmware "mtd -r write /tmp/openwrt-brcm47xx-squashfs.trx linux" Mar 23 16:48:01 crow: oh, well, then do that Mar 23 16:48:10 but never from usb stick openwrt :) Mar 23 16:48:25 crow: it doesn't matter Mar 23 16:48:33 hope it doesnt matter :) Mar 23 16:48:57 cshore1 ok thnx well I will wait for fix and then reflash it and test it :) Mar 23 16:49:03 ok Mar 23 16:59:29 http://openwrt.pastebin.ca/1850520 should fix the problem Mar 23 17:02:03 basically jffs is mounted on /tmp/overlay before the extroot is mounted. If there is an extroot it is mounted on /overlay, and we check for that. Unfortunately the check was always succeeding because it was taking the /overlay part of /tmp/overlay as being successful mount...I've changed this to only treat exactly /overlay as success Mar 23 17:02:59 xMff: could you commit this, please? Mar 23 17:03:18 was about to ask Mar 23 17:05:01 grr Mar 23 17:05:19 pc at work can reach pastebin.ca but not commit, pc at home can commit but not reach pastebin Mar 23 17:05:53 well that sucks....do you want me to try a different pastebin? Mar 23 17:06:35 or email the list? Mar 23 17:06:36 no, its okay Mar 23 17:06:41 scp ftw Mar 23 17:07:11 I'm just used to do GET http://blah | patch -pN Mar 23 17:08:39 jow * r20390 /trunk/package/block-extroot/files/50_determine_usb_root: [package] block-extroot: make overlay mount check more explicit, it wrongly detected /tmp/overlay as successful overlay mount Mar 23 17:15:20 any idea what could be overwriting /etc/hosts during/after sysupgrade? I explicitly enabled it in config 'extern' 'flash_keep' in /e/c/luci Mar 23 17:16:19 stintel: is /etc/hosts defined as something sysupgrade is supposed to keep? (/lib/upgrade/common.sh) Mar 23 17:16:43 thepeople: I defined it in /etc/config/luci to keep it Mar 23 17:16:48 stintel: flash_keep is not used by sysupgrade, only by luci-flash (which is used by the ui) Mar 23 17:17:00 ah, no kidding :/ Mar 23 17:19:32 stintel: isn't /etc/hosts a symlink to /tmp/etc/hosts Mar 23 17:19:40 cshore1: not anymore Mar 23 17:20:21 cshore1: I got rid of luci_{hosts,ethers} + the symlinks and added migration code to put the stuff into /etc/config/dhcp now on firstboot Mar 23 17:21:04 well I need /etc/hosts since I have some machines with static IP. Mar 23 17:21:43 so dnsmasq doesn't use /etc/hosts anymore (or only for manually defined routes)? Mar 23 17:21:45 defining the hostnames in luci/dnsmasq only works if the machine requested IP via DHCP Mar 23 17:21:49 maybe we should add /etc/hosts and /etc/ethers to the default sysupgrade list now that they're not mangled by the ui anymore Mar 23 17:22:23 stintel, cshore1 no, there's an extra facility to define dhcp-independant hostnames Mar 23 17:22:37 stintel: you can use a domain directive Mar 23 17:22:46 http://wiki.openwrt.org/doc/uci/dhcp#custom.domain Mar 23 17:23:06 ah Mar 23 17:23:08 interesting Mar 23 17:24:15 jow: perhaps the same thing as domain should be done automatically for dhcp-defined hostnames (otherwise you have to enter the host and ip twice) Mar 23 17:24:30 ok please close #6904 invalid :) Mar 23 17:26:46 hmmm it would be nice if config 'host' took both things into account Mar 23 17:26:56 if possible Mar 23 17:29:09 but the option 'domain' works, and I like it better than putting all that in /etc/hosts Mar 23 17:29:37 cshore1: what do you mean with same thing? the ptr record? Mar 23 17:29:47 yes Mar 23 17:30:08 it needs some work anyway because it does not work with ipv6 addresses Mar 23 17:30:33 and we have no suitable utility in the base system to resolve ipv6 addresses to their canonical format Mar 23 17:30:59 e.g. ::1 -> 0:0:0:0:...:0:1 Mar 23 17:31:28 or f00:ba3::ba2:1.2.3.4 Mar 23 17:32:09 anyway Mar 23 17:32:44 doesn't dnsmasq does the ptr records by itself for dhcp leases? Mar 23 17:38:57 cshore1 ok thnx for that fix as soon as it in svn i will compile and try. Mar 23 17:39:23 ok its already inside :) Mar 23 17:45:03 cshore1 so if i put "is_rootfs 0" to /etc/config/fstab inside usb stick openwrt and do reboot it should boot to flash? or this way just work if no usb stick? Mar 23 17:46:20 or i need to put that inside /etc/config/fstab which on flash is? but then if on usb stick its readonly /rom mount point Mar 23 18:26:38 nico * r20391 /packages/net/sslcat/ (. Makefile): [packages] add 'sslcat' package Mar 23 18:45:54 stintel: do you know if the patches will work with wpad? Mar 23 18:46:14 thepeople: they do Mar 23 18:46:16 thepeople: I use wpad Mar 23 18:46:29 ok :-) Mar 23 18:47:35 crow: no, you have to specifiy the is_rootfs 0 on the jffs...if you are in the usb root then you do mount -t jffs2 /dev/mtdblock3 /mnt (it might be mtdblock4) Mar 23 18:48:20 stintel: they no longer apply cleanly, can you send me a new patch file Mar 23 18:48:25 cat /proc/mtd to see Mar 23 18:48:32 stintel: you can do it all in one if you want Mar 23 18:49:22 thepeople: hold on Mar 23 18:50:38 stintel: no hurries :-) Mar 23 18:52:02 thepeople: to you directly ? Mar 23 18:53:16 stintel: sure, or drop it in a pastebin Mar 23 19:00:30 thepeople: sent :) Mar 23 19:09:17 stintel: alright I will look for it Mar 23 19:11:04 cshore1 will you add block-extroot to luci ?` Mar 23 19:11:20 someday Mar 23 19:12:05 hmm, would be a cool thing to have this in backfire :) Mar 23 19:17:16 cshore1 i have rootfs and rootfs_data http://openwrt.pastebin.com/HBjrXGiB Mar 23 19:17:45 you want rootfs_data Mar 23 19:17:54 ok Mar 23 19:21:10 cshore1 which mean if I change it now and reboot it should boot flash even that fix isnt included? Mar 23 19:21:49 if you change /mnt/etc/config/fstab, yes Mar 23 19:23:12 :) Mar 23 19:23:37 maybe that should be also added to wiki "Debug info or something" Mar 23 19:28:02 perhaps I should have asked this beforehand; which is the preferred way of sending patches to openwrt-devel: inline or as attachment? Mar 23 19:29:52 ah, just found it Mar 23 19:35:49 maybe that should be also added to wiki "Debug info or something" /2 (you quite before i posted it) Mar 23 19:36:19 crow: sorry, what's ?that Mar 23 19:37:19 crow: you can edit the wiki you know, you just have to sign up for an account Mar 23 19:37:50 cshore that about mounting jffs2 from flash to edit fstab and reboot without puting usb stick out (ideal for remote debuging).. Mar 23 19:38:03 cshore yea i will do that for shure. Mar 23 19:38:22 crow: thanks...only so many brains that I have you know Mar 23 19:39:12 crow: that darn, one of me in one place thing Mar 23 19:57:42 stintel: did your rebuild finish yet? Mar 23 19:57:55 thepeople: swconfig ? Mar 23 19:57:55 wondering about swconfig Mar 23 19:58:03 yep :-) Mar 23 19:58:06 thepeople: mea culpa, forgot to feedback Mar 23 19:58:07 works fine Mar 23 19:58:36 good to hear Mar 23 19:59:59 stintel: was the luci issue fixed as well? Mar 23 20:00:55 jow_laptop: yes, r5992 fixed it Mar 23 20:01:05 thanks again Mar 23 21:38:27 mb here? Mar 23 22:06:07 ping cshore Mar 23 22:08:10 xMff: how was the dir called to execute scripts once? Mar 23 22:08:17 xMff: etc/uci-defaults? Mar 23 22:08:31 yep Mar 23 22:12:30 xMff: thanks Mar 23 22:24:43 does anybody here now where and when the timer irq gets enabled on mips systems? Mar 23 22:29:09 rtz2: i think linux expects it to be enabled by the bootloader; at least that was the case on bcm47xx Mar 23 22:30:55 pong xMff Mar 23 22:30:59 KanjiMonster: you really sure about that? Mar 23 22:31:29 KanjiMonster: because the platfrom irq dispatcher is called the first time shortly after cpu initialization Mar 23 22:31:31 cshore: the failsafe message in the udp broadcast is chopped off, wonder if you're aware of this? Mar 23 22:31:48 xMff: no I wasn't Mar 23 22:31:49 cshore: it only contains "Please" Mar 23 22:32:33 and on x86 I see "Press Press f to enter failsafe mode" Mar 23 22:32:33 xMff: hmmm....obviously my keypress thingy isn't doing the network echo right Mar 23 22:32:36 maybe related? Mar 23 22:33:05 (and it never worked with enter :P) Mar 23 22:33:22 xMff ah i see that to now :) Mar 23 22:33:23 rtz2: I tried writing an bcm47xx emulator once, and linux needed the timer irq enabled (at least the mask bit set) when calculating the bogomips Mar 23 22:34:09 but I guess this can be different on other mips machines Mar 23 22:34:53 KanjiMonster: it got till "calibrating delay loop ..." and simply hang there? Mar 23 22:35:10 cshore: I have to admit that I did not really looked into it, maybe you have an idea already Mar 23 22:35:45 rtz2: thats what I saw until I made the cfe enable the bit for the timer irq Mar 23 22:36:18 KanjiMonster: ok, I have exactly the same problem Mar 23 22:36:51 rtz2: but my problem was that the emulator wasn't good enough, not on real hardware - so I doubt this helps Mar 23 22:37:11 KanjiMonster: it may do Mar 23 22:37:43 KanjiMonster: which bit did you enable exactly? CAUSEF_IP7 in c0 status? Mar 23 22:37:49 xMff: well i'm looking at it now and I don't understand the symptoms Mar 23 22:37:59 they make no sense Mar 23 22:38:28 rtz2: that was the bit that needed to be set Mar 23 22:39:06 KanjiMonster: I tried to set it, but then the kernel dies somewhere in the irq handling code Mar 23 22:39:26 KanjiMonster: let me try to set it somewhere really early Mar 23 22:39:35 rtz2: what platform is it? Mar 23 22:40:18 KanjiMonster: bcm4716 Mar 23 22:41:09 ah Mar 23 22:41:25 cshore: does "press f" mean that I am supposed to press ? Mar 23 22:41:44 mip32r2 iirc Mar 23 22:42:32 KanjiMonster: right Mar 23 22:42:55 KanjiMonster: i'm using it in r1 mode, would this be a problem? Mar 23 22:42:57 xMff: yes Mar 23 22:43:03 ahh Mar 23 22:43:13 I always thought it is just a typo :P Mar 23 22:43:43 which mean to type: f and then pres enter? Mar 23 22:43:43 rtz2: I only tried to emulate an r1, but as far as I have seen the mips documentation, it should behave the same Mar 23 22:43:56 xMff: yes Mar 23 22:44:03 got it now, and it works Mar 23 22:44:18 * xMff always hit enter and wondered why it didn't work Mar 23 22:44:39 heh...never thought of that Mar 23 22:45:15 i was thinking the same, trough never tried it Mar 23 22:45:27 xMff: it also took me some time to figure it out Mar 23 22:45:27 http://openwrt.pastebin.com/LEdirvXg Mar 23 22:45:31 "type and pres enter" should be better Mar 23 22:46:15 ^ Mar 23 22:46:26 or both need to be pressed in same time? Mar 23 22:46:34 one after another is fine Mar 23 22:47:08 it's the little things that get you Mar 23 22:47:35 then "Type and press " is for me user frendly :) Mar 23 22:47:49 xMff: I'm happy with whatever people think will work Mar 23 22:48:40 * xMff bets there'll be people typing [<] [f] [>] [enter] Mar 23 22:48:50 * cshore agrees Mar 23 22:49:19 * cshore thinks there is not perfect solution Mar 23 22:49:40 "Press the <$1> key and hit $2" Mar 23 22:50:07 xMff: that's very good I think Mar 23 22:50:11 ^ is right one :) Mar 23 22:50:23 "Press the [$1] key and hit [enter] $2" Mar 23 22:50:28 probably better Mar 23 22:51:34 rtz2: I guess you are trying to get the wnr3500 to work? ;) Mar 23 22:51:59 now I only need to figure out where the redundant "Press" comes from Mar 23 22:52:36 Anyone know why preinit_net_echo "Please press button now to enter failsafe" would only print Please? Mar 23 22:53:03 KanjiMonster: asus ar-n16 Mar 23 22:53:07 it just a call to netmsg (essentially) netmsg broadcast "$1" Mar 23 22:53:25 cshore: maybe the sentence is improperly quoted when passed to the udpsend program so it gets separated into multiple arguments Mar 23 22:53:28 cshore: try $* or something Mar 23 22:54:02 KanjiMonster: rt-n16 Mar 23 22:54:24 rtz2: ah, ok Mar 23 22:54:46 KanjiMonster: got one from glp Mar 23 22:55:12 rtz2: have you checked how the asus stock firmware handles it? Mar 23 22:55:47 KanjiMonster: the dd-wrt kernel boots, but I have no clue, how they avoid this problem Mar 23 22:55:58 KanjiMonster: but the timer setup is a little bit different Mar 23 22:56:15 jow * r20392 /trunk/package/base-files/files/lib/preinit/30_failsafe_wait: [package] base-files: make press key message more explicit in preinit Mar 23 22:57:17 jow * r20393 /trunk/target/linux/x86/base-files/lib/preinit/45_failsafe_x86: [x86] base-files: remove redundant "Press " when writing enter failsafe message Mar 23 22:57:40 KanjiMonster: http://svn.dd-wrt.com:8000/dd-wrt/browser/src/linux/brcm/linux-2.6.23/arch/mips/brcm-boards/bcm947xx/time.c#L182 Mar 23 22:57:43 ah, a different target file Mar 23 22:58:19 KanjiMonster: but I have no idea, how this makes a difference Mar 23 22:58:40 KanjiMonster: it's not needed for all the other brcm cpus Mar 23 22:59:53 rtz2: as far as i see bcm4716/18 is a quite a different architecture as the earlier cpus Mar 23 22:59:58 xMff: http://openwrt.pastebin.ca/1850884 Mar 23 23:00:13 xMff: maybe that will solve the udp broadcast? Mar 23 23:00:24 KanjiMonster: yes, but this is generic mips stuff Mar 23 23:00:30 cshore: can you put it in pastebin.com please? Mar 23 23:00:40 .ca is unreachable again Mar 23 23:00:43 for me Mar 23 23:01:15 KanjiMonster: the counter/compare register stuff and irq 7 is part of the mips32 architecture Mar 23 23:02:05 xMff: http://openwrt.pastebin.com/3VxDLAYG Mar 23 23:02:38 rtz2: but it might depend on how the cfe sets up the cpu - I think asus is nice and provides the cfe sources Mar 23 23:10:01 build #39 of octeon is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/octeon/builds/39 Mar 23 23:15:35 bahhh Mar 23 23:15:47 I have no clue, how this clockevent stuff is supposed to work Mar 23 23:34:57 KanjiMonster: it was another problem Mar 23 23:35:14 KanjiMonster: it used the wrong irq Mar 23 23:36:50 hi there Mar 23 23:37:49 o/ Mar 23 23:37:54 nbd: hello Mar 23 23:38:33 nbd: do you have some time? I have to pick you brain about brcm47xx cacheline stuff :) Mar 23 23:38:44 not much time Mar 23 23:38:55 i'm still at atheros Mar 23 23:39:21 and i will leave in a few minutes Mar 23 23:40:06 nbd: this patch breaks for the brcm4716: https://dev.openwrt.org/browser/trunk/target/linux/brcm47xx/patches-2.6.32/150-cpu_fixes.patch Mar 23 23:40:11 nbd: no problem Mar 23 23:40:28 nbd: I fixed it for now, a closer look at it can wait Mar 23 23:41:13 how did you fix it? Mar 23 23:43:06 nbd: by removing __BUILD_BLAST_CACHE_RANGE(d, dcache, Hit_Writeback_Inv_D, protected_, BCM4710_PROTECTED_FILL_TLB(addr); BCM4710_PROTECTED_FILL_TLB(aend);, BCM4710_DUMMY_RREG();) Mar 23 23:43:30 is it just the tlb fill that's causing issues? Mar 23 23:43:32 or anything else as well? Mar 23 23:43:34 nbd: I did remove the bcm stuff in it, not the complete line Mar 23 23:44:09 nbd: not sure, I run into other problems pretty soon afterwards Mar 23 23:44:46 what was the symptom of the breakage? Mar 23 23:44:47 a crash? Mar 23 23:44:49 nbd: I will have to see, once I get a booting kernel Mar 23 23:45:10 nbd: not 100% sure, but I think the kernel silently died Mar 23 23:45:16 ok Mar 23 23:45:32 ah yes Mar 23 23:45:35 i know why this happens Mar 23 23:45:45 i think some newer stuff has a different system bus Mar 23 23:45:55 that is not entirely compatible with the old ssb stuff Mar 23 23:46:06 nbd: yes, I already noticed Mar 23 23:46:13 and at least one of these macros does a backplane read to flush the tlb Mar 23 23:47:28 nbd: right, I think I found it Mar 23 23:47:42 so a cpu rev check in this backplane read should take care of it Mar 23 23:47:53 imho the extra memory access for the cpu rev check shouldn't be a problem Mar 23 23:47:54 nbd: a bunch of registers are different on the 16 Mar 23 23:48:09 and if it is a problem, we can make it depend on a cp0 register access in an inline function Mar 23 23:48:33 nbd: does it have to be a specific reg? Mar 23 23:48:41 i don't know Mar 23 23:49:05 nbd: if not, we could switch it to a common reg Mar 23 23:49:15 BCM4710_DUMMY_RREG doesn't suggest so Mar 23 23:49:22 nbd: well, we will see Mar 23 23:52:25 nbd: why are those there in the first place? Mar 24 00:19:45 thepeople * r20394 /trunk/package/hostapd/files/wpa_supplicant.sh: Mar 24 00:19:45 Cleanup redundant psk handling code in wpa_supplicant.sh. Mar 24 00:19:45 (I sent this one to openwrt-devel before, but I am resending it as part of Mar 24 00:19:45 this series because the next patch will also edit wpa_supplicant.sh). Mar 24 00:19:45 Signed-off-by: Stijn Tintel Mar 24 00:21:30 thepeople * r20395 /trunk/package/hostapd/files/ (wpa_supplicant-full.config wpa_supplicant.sh): Mar 24 00:21:30 Enable management frame protection in wpa_supplicant, and make it Mar 24 00:21:30 configurable in /etc/config/wireless. Mar 24 00:21:30 Signed-off-by: Stijn Tintel Mar 24 00:22:24 thepeople * r20396 /trunk/package/hostapd/ (Makefile files/hostapd.sh): Mar 24 00:22:25 Enable management frame protection in hostapd, and make it configurable Mar 24 00:22:25 in /etc/config/wireless. Since ath9k is currently the only driver that Mar 24 00:22:25 supports MFP, it will only be enabled when ath9k is enabled. Mar 24 00:22:25 Signed-off-by: Stijn Tintel Mar 24 00:23:12 thepeople * r20397 /trunk/package/hostapd/files/hostapd.sh: Mar 24 00:23:12 When enabling MFP, hostapd will read assoc_sa_query_max_timeout and Mar 24 00:23:12 assoc_sa_query_retry_timeout from it's config file. Make these options Mar 24 00:23:12 configurable in /etc/config/wireless. To make it clear that these Mar 24 00:23:12 options are 802.11w related, I named them ieee80211w_max_timeout and Mar 24 00:23:12 ieee80211w_retry_timeout instead. Mar 24 00:23:13 Signed-off-by: Stijn Tintel Mar 24 00:23:26 \kick CIA-39 flood! :-) Mar 24 00:23:27 stintel: ^ Mar 24 00:23:33 thepeople: cheers Mar 24 00:34:28 builder: show Mar 24 00:34:36 builder: help Mar 24 00:34:36 Get help on what? (try 'help ', or 'commands' for a command list) Mar 24 00:34:44 builder: commands Mar 24 00:34:44 buildbot commands: commands, dance, destroy, excited, force, hello, help, join, last, leave, list, notify, source, status, stop, version, watch Mar 24 00:34:53 builder list Mar 24 00:34:57 builder: list Mar 24 00:34:57 try 'list builders' Mar 24 00:35:03 builder: status Mar 24 00:35:03 adm5120: idle, last build 28h39m27s ago: failed compile_4 Mar 24 00:35:03 ar7: building(compiling packages) [ETA 22h18m55s] Mar 24 00:35:03 ar71xx: idle, last build 23h11m37s ago: failed compile_4 Mar 24 00:35:03 atheros: building(compiling packages) [ETA 4h39m39s] Mar 24 00:35:03 au1000: idle, last build 27h09m09s ago: failed compile_4 Mar 24 00:35:16 stintel: that's how it's done :D Mar 24 00:35:43 rtz2: hmmm? Mar 24 00:35:58 the patches? Mar 24 00:37:00 builder: status Mar 24 00:37:01 adm5120: idle, last build 28h41m25s ago: failed compile_4 Mar 24 00:37:01 ar7: building(compiling packages) [ETA 22h18m38s] Mar 24 00:37:01 ar71xx: idle, last build 23h13m35s ago: failed compile_4 Mar 24 00:37:01 atheros: building(compiling packages) [ETA 4h37m25s] Mar 24 00:37:01 au1000: idle, last build 27h11m07s ago: failed compile_4 Mar 24 00:37:20 stintel: kicking bots Mar 24 00:37:24 :D Mar 24 00:37:36 I am on irc for too long Mar 24 00:37:50 sometimes it's a pity that I don't have ircop status here :) Mar 24 00:39:58 rtz2: picking on the poor builder bot Mar 24 00:40:21 builder: destroy rtz2 Mar 24 00:40:21 * builder readies phasers Mar 24 00:40:45 stintel: can you please update doc/uci/wireless in the wiki after your patches? Mar 24 00:47:45 xMff: I'll have a look at it Mar 24 00:48:39 xMff: I need to login, I assume ? Mar 24 00:48:47 yes Mar 24 00:48:52 where's that noob hat again :) Mar 24 00:49:07 * xMff give stintel a noob hat Mar 24 00:49:34 * xMff gives stintel a wiki gardeners toolbox as well Mar 24 00:49:38 hf Mar 24 01:04:07 lol, good there is a preview button. I fail :) Mar 24 01:04:58 * thepeople grabs the jumbo sized noob hat for stintel Mar 24 01:07:27 hrrm, should they go in the table of common options or rather WPA Enterprise ? Mar 24 01:09:47 oh well, I put them under common Mar 24 01:17:42 thanks Mar 24 01:19:42 yw, now if I figure out how to let buildroot apply a patch to luci (without failing), I can send patches for that as well :) Mar 24 01:19:55 I tried but failed, wrong -pN option to patch, I believe Mar 24 01:20:17 it is a bit tricky since it downloads its own source Mar 24 01:20:22 yeah Mar 24 01:20:48 I'd like to add the holdoff and maxfail options to pppd config as well Mar 24 01:20:54 if you checkout it separetly and link that through feeds it should work Mar 24 01:21:33 src-link luci /home/foo/luci/{trunk,branches/luci-0.9}/contrib/package Mar 24 01:22:31 patches preferred for trunk? Mar 24 01:22:58 yes Mar 24 01:23:51 ok, should I expect any problems running luci trunk ? if yes I'd better finish the x86-kvm/virtio stuff first, allows for way easier testing than using my home router :P Mar 24 01:24:28 yes, you'll see some broken stuff in trunk but it'll be in a runnable state Mar 24 01:24:37 it is in the middle of an ui rewrite now Mar 24 01:25:16 ok, better test it on something else then, just to be sure. could avoid me pulling my hair. Mar 24 01:25:24 yep Mar 24 01:25:34 I'd consider the kvm target more important Mar 24 01:25:35 maybe the noob hat helps against that as well ;-) Mar 24 01:26:03 yeah, I'll finish that first Mar 24 01:28:42 hrrm, was the patch that adds rtc support for the rspro ever looked at? Mar 24 01:29:27 no, was that yours? Mar 24 01:29:35 (wasn't my patch but I remember I saw it passing on openwrt-devel someday) Mar 24 01:29:45 ah Mar 24 01:29:59 I still havent flashed my rspro :P Mar 24 01:30:43 I've been using mine for months, but it's only starting to become near fully usable recently Mar 24 01:31:07 the "support" from ubnt for the rspro was a joke imo Mar 24 01:31:12 yeah Mar 24 01:31:22 they totally relied on 3rd parties to sort the stuff out Mar 24 01:31:57 xMff: they have never been much help with porting to any of their devices Mar 24 01:32:13 it would actually be nice if ar71xx and ag71xx stuff could go upstream, might receive more love from other developers then Mar 24 01:32:16 they haven't even ported their own os to that board and then they offered their 900mhz cards without any driver Mar 24 01:32:19 or am I dreaming now :) Mar 24 01:33:02 xMff: well they have gotten away with it Mar 24 01:33:08 it seems to work for them Mar 24 01:33:14 yeah Mar 24 01:33:26 likely helps they have good hardware :-) Mar 24 01:33:35 I mean for the proce point it is probably okay Mar 24 01:33:36 well even though their linux support for it is pretty bad, the hardware really is nice, imho Mar 24 01:33:42 *price Mar 24 01:33:51 at least they released their SDK Mar 24 01:34:01 OutBackDingo: but even that is always late Mar 24 01:34:03 and the price of rspro compared to ar71xx based rb is quite cheap Mar 24 01:35:09 true but the rb ships with a usable OS at least :) Mar 24 01:35:34 ah right, forgot about that, likely that's what explains the price difference Mar 24 01:35:40 ;) Mar 24 01:35:49 damn, my ISP started having packet loss few minutes ago Mar 24 01:36:04 well, getting an os in a weel working state is a lot of work Mar 24 01:36:12 as everybody here can testify Mar 24 01:36:13 that means bedtime, irc over ssh over laggy connection is big PITA Mar 24 01:36:25 good timezone / happy hacking all :) Mar 24 01:36:32 UGT stintel Mar 24 01:48:10 xMff: did you also have a device with a brcm4716 cpu? Mar 24 01:48:28 xMff: asus rt-n12 or something? Mar 24 01:50:51 I have an asus n-10 with a chip labelled 5356a1kfbg but I could not find anything about it Mar 24 01:50:58 *rt-n10 Mar 24 01:52:07 xMff: did you try to get openwrt running on it? Mar 24 01:52:34 a while ago but I fried cfe and haven't gotten around to re-jtag it yet Mar 24 01:54:09 xMff: hmm, ok Mar 24 01:54:27 xMff: do you know of any other devices with the 4716? Mar 24 01:54:37 xMff: it should have usb and wireless n mimo Mar 24 01:54:51 xMff: and maybe pci-e Mar 24 01:55:58 wikipedia lists several Mar 24 01:56:15 it does? Mar 24 01:56:17 let me check Mar 24 01:56:28 http://en.wikipedia.org/wiki/Linksys_routers Mar 24 01:56:36 search for 4716 Mar 24 01:59:15 xMff: ok, thanks Mar 24 01:59:42 you might be able to steal some stuff from dd-wrt Mar 24 01:59:54 hmm, If I get this working, this could add support for some of the wrt160n models Mar 24 01:59:58 *borrow :P Mar 24 02:00:04 there were some people asking for it Mar 24 02:00:07 not really Mar 24 02:00:19 xMff: the ssb stuff is completly different Mar 24 02:00:24 they're still on 2.4 ? Mar 24 02:00:40 no, but they use broadcom code Mar 24 02:00:47 ah Mar 24 02:01:03 it looks, like it runs on 2.4, 2.6 and even non-linux OSs Mar 24 02:01:16 there are at least some defines suggesting it Mar 24 02:01:21 yeah Mar 24 02:01:30 took a look at their wl.ko a while ago Mar 24 02:02:13 the large-scale structure is similiar, but the details are completly different **** ENDING LOGGING AT Wed Mar 24 02:59:57 2010