**** BEGIN LOGGING AT Wed Feb 15 02:59:57 2012 Feb 15 06:03:41 build #119 of x86 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/119 Feb 15 06:25:49 blogic: is anyone really still using slip? Feb 15 07:17:20 juhosg * r30547 /trunk/target/linux/ramips/ (5 files in 4 dirs): ramips: raeth: add PHYLIB support Feb 15 08:20:01 build #123 of rb532 is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/123 Feb 15 08:48:09 build #113 of avr32 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/113 Feb 15 09:33:23 hi all. I'm running OpenWRT trunk on D-Link DIR-330, and its WAN port (assuming its eth1) does not work. I googled, wikied and so on, and nothing helped so far. Feb 15 09:34:02 I've found a patch which forces eth1 to be on the same phy 30 as eth0, but mine are already on the same phy. or is this a problem? Feb 15 09:35:56 I also found a patch that by adding modified defaults into netconfig makes some magic with switch on eth1, but that didn't work either Feb 15 09:37:05 <_trine> I have just flashed one of my Dockstars but when I look in LUCI it displays the router model as Marvell SheevaPlug Reference Board Feb 15 09:37:34 <_trine> Firmware Version Feb 15 09:37:34 <_trine> OpenWrt Firmware Attitude Adjustment (r30508) / Feb 15 09:37:34 <_trine> LuCI Trunk (trunk+svn8224) Feb 15 09:37:34 <_trine> Feb 15 09:41:50 okay, I'll try to simplify my question. Is the fact that eth0 and eth1 both are on phy 30 normal? Feb 15 09:43:38 Barvinok: i don't have a d-link dir-330, so i have no idea ;-) Feb 15 09:44:27 so at least somebody is alive here :) well then, what steps should I undertake to diagnose my problem? Feb 15 09:50:13 ok then. another question: my dmesg is getting flooded with "hub 1-0:1.0: unable to enumerate USB device on port 2" few times a second. how do I diagnose that? Feb 15 09:50:41 try rmmod'ing the usb module? Feb 15 09:51:07 i know the wgt634u didn't like kmod-usb-uhci or something Feb 15 09:51:41 lsmod should tell you want is loaded Feb 15 09:52:40 well... the problem is I didn't use modules. my kernel was build monolythic Feb 15 09:52:49 so lsmod gives nothing Feb 15 09:53:47 disabling this uhci or ohci and reflashing isn't a problem, but will my USB work after that? Feb 15 10:00:46 Barvinok: in my case, one of the modules was throwing errors, the other one worked Feb 15 10:01:06 Barvinok: is that the default config for your target? Feb 15 10:01:14 aha, ok. will try using either one of them alone Feb 15 10:01:35 no it's not default. I ran menuconfig and screwed up much Feb 15 10:01:56 you can always rm .config .config.old and start over ;-) Feb 15 10:02:07 no, I mean it's working config. Feb 15 10:02:34 the usb is just annoying. the only serious problem is eth1 Feb 15 10:03:24 i know nothing about your device, so /me just guesses ... i'd recommend you continue lurking until someone who knows wakes up Feb 15 10:03:43 it's plain old brcm47xx Feb 15 10:04:00 is it supported? Feb 15 10:04:07 officially? Feb 15 10:04:12 how do I know? Feb 15 10:04:16 cshore * r30548 /packages/ (ipv6/ndppd/Makefile net/kismet/Makefile net/rrs/Makefile): [packages] more gcc 4.6 compilation fixes Feb 15 10:04:33 look at the supported devices table linked on the home page Feb 15 10:05:18 http://wiki.openwrt.org/toh/start Feb 15 10:05:34 i don't see it. but I do see DIR-320 there, which only differs by amount of flash memory Feb 15 10:05:53 and there are also http://wiki.openwrt.org/toh/d-link/dir-330 Feb 15 10:06:18 So though not 'officially', I believe it is supported as compatible platform Feb 15 10:08:48 does it get detected during boot? Feb 15 10:08:56 yes Feb 15 10:09:03 as what? Feb 15 10:09:14 dmesg says [ 5.320000] diag: Detected 'D-Link DIR-330' Feb 15 10:09:19 ah Feb 15 10:09:33 well, that's a good sign ;-) Feb 15 10:09:37 should I post complete dmesg ? Feb 15 10:09:58 not for me Feb 15 10:10:50 well then. of 94 other people here, what are the timeframe of maximum chance of getting relevant suggestions? Feb 15 10:11:13 i would ask Hauke, he's the brcm47xx guy Feb 15 10:11:35 what timezone are you in? Feb 15 10:11:56 utc+2 Feb 15 10:12:18 ouch, you should be asleep! Feb 15 10:12:32 no i shouldn't. it's 12am here Feb 15 10:12:37 midday Feb 15 10:13:10 i'm usually asleep then ;-) Feb 15 10:15:58 I am currently reading the backlog Feb 15 10:16:28 dmesg - http://pastebin.com/rjsvXUEQ Feb 15 10:16:58 for it i got usb disabled Feb 15 10:20:58 Barvinok: "SIOCGETCPHYRD failed!" doesn't look happy Feb 15 10:25:12 hmm, something is wrong in the switch driver, but I do not have worked much on the switch driver Feb 15 10:29:17 is it Ok for eth0 and eth1 to be on the same phy? Feb 15 10:29:27 'cause I don't really know what phy is... Feb 15 10:30:47 <[florian]> a PHY is the chip which basically transforms an ethernet frame into bits on the wire Feb 15 10:31:01 <[florian]> eth0 and eth1 can be either using different PHYs Feb 15 10:31:04 <[florian]> they are independant Feb 15 10:31:23 <[florian]> but usually you have one ethN interface and a switch Feb 15 10:31:38 <[florian]> which presents itself like a "special" PHY to the ethernet adapter Feb 15 10:32:29 wait... it's not the dmesg from the firmware I built. sorry Feb 15 10:32:35 I'll repost in a moment Feb 15 10:35:19 I will have lunch now Feb 15 10:41:12 here is the new dmesg, this time from the fw I built -- http://pastebin.com/Bv23E3KL Feb 15 10:41:34 it's a bit weirder, reporting kernel deadlock or something Feb 15 10:42:26 I wonder why robocfg says that switch is on eth1 Feb 15 11:31:45 guys, please commit my patches for packages/ipv6/taygs... Feb 15 11:34:51 <_trine> There appears to be a problem downloading this file while I'm compiling trunk Feb 15 11:34:57 <_trine> Download failed. Feb 15 11:34:57 <_trine> --2012-02-15 11:32:23-- http://downloads.openwrt.org/sources/zeromq-2.1.9.tar.gz Feb 15 11:35:42 <_trine> Download failed.No more mirrors to try - giving up. Feb 15 11:39:35 <_trine> probably wants updating to 2.1.11 Feb 15 11:48:45 is there structural differences between broadcom ref firmware and openwrt firmware ? Feb 15 11:48:55 ie order of data Feb 15 11:49:21 I have trouble getting cfe to flash openwrt generated images properly Feb 15 12:09:57 merbanan: depends on the device, some vendors are using some special header in their image Feb 15 12:10:26 I am a "vendor" Feb 15 12:12:43 just wondering if the openwrt firmware are supposed to be flashable from cfe Feb 15 12:13:17 or do I need to flash from the vendor firmware Feb 15 12:13:31 merbanan: the openwrt-brcm47xx-squashfs.trx image is flashable from cfe Feb 15 12:14:30 at lest if you or brodacom haven't changes something recently Feb 15 12:15:45 I am using imagetag to generate the images Feb 15 12:17:12 trx images seem to use some other tool Feb 15 12:20:10 merbanan: are working on brcm63xx or brcm47xx ? Feb 15 12:20:22 6368 Feb 15 12:20:52 <[florian]> merbanan: what kind of issues do you have exactly? Feb 15 12:21:22 the thing is a coworker patches the boot loader to make some openwrt images to boot Feb 15 12:21:46 merbanan: ah, I do not have much knowledge about the brcm63xx target, but [florian] is the expert Feb 15 12:22:26 that works but wont help other targets where the bootloader cant be changed Feb 15 12:22:54 so I'd rather figure out how to generate proper images Feb 15 12:23:59 the image tagging and crc calculation works Feb 15 12:24:30 it's just the offsets and image structures i don't understand Feb 15 12:26:54 image structure -> flash layout Feb 15 12:27:06 openwrt vs broadcom ref Feb 15 12:42:18 <[florian]> KanjiMonster: do you remember if we generate broadcom compatible imagetag images now? Feb 15 12:43:31 [florian]: not sure what you mean with "broadcom compatible" Feb 15 12:44:11 <[florian]> KanjiMonster: I mean the same layout (kernel/rootfs order) as what their tool generates Feb 15 12:44:41 [florian]: no we don't Feb 15 12:46:00 merbanan: maybe this (not yet submitted ustream) patch clears things a bit up -> https://github.com/KanjiMonster/bcm63xx/commit/10be940cd518a9419aebc8634a9362cbc7106c76 Feb 15 12:47:27 merbanan: but generally openwrt images are supposed to be flashable from CFE (and they work for all devices I have) Feb 15 12:48:23 KanjiMonster: right now I don't even get the cfe to boot the kernel Feb 15 12:48:38 but thanks I'll look into the patch Feb 15 12:48:53 not sure it relates to my issues Feb 15 12:48:56 <[florian]> merbanan: we have not yet integrated the bcm6368 patches into openwrt Feb 15 12:49:00 <[florian]> but that's planned of course Feb 15 12:49:17 merbanan: it doesn't, it just "explains" the difference between openwrt and broadcom images Feb 15 12:50:01 ok, so the kernel and filesystem are switched ? Feb 15 12:50:06 merbanan: if you are using trunk, you need to enable bcm6368 support in target/linux/brcm63xx/config-3.3 (it's currently disabled since it isn't really practical yet, the ethernet support needs to be added) Feb 15 12:50:41 KanjiMonster: I'm using a custom kernel :/, need the enpoint support Feb 15 12:51:00 endpoint Feb 15 12:51:16 <[florian]> endpoint is not very complicated to add back to a vanilla kernel Feb 15 12:51:27 <[florian]> but you might also need the bcm hal interrupt crap Feb 15 12:54:14 merbanan: assuming you have access to the endpoint driver sources it should also be relatively easy to rewrite it to not use the bcm hal interrupt crap ;) (iirc the endpoint driver was rather small) Feb 15 12:54:29 well, the part that interacted with linux ;) Feb 15 12:55:56 good luck getting broadcom support for that :) Feb 15 12:56:06 <[florian]> we know Feb 15 12:56:38 anyway, the broadcom 4.12 release has some new gpl switch source code, have you looked at it ? Feb 15 12:56:48 <[florian]> yes Feb 15 12:58:10 but porting the endpoint driver sounds like an interesting idea Feb 15 12:58:24 broadcom said they want to gpl most of the code in the broadcom SDK Feb 15 12:58:37 what is the release date of this sdk? Feb 15 12:58:44 Hauke: different division ;) Feb 15 12:58:45 <[florian]> Hauke: I guess this is true for bcm47xx, for bcm63xx there are still quite a lot of binary code out tehre Feb 15 12:59:24 Hauke: 2012-01-11 Feb 15 12:59:36 merbanan: efixo did something like that for their neufboxes (using openwrt as a base and the broadcom drivers on top) Feb 15 12:59:37 I just talked to Arned how is in an other different devision ;-) Feb 15 12:59:44 *who Feb 15 13:00:00 *Arend ;p Feb 15 13:00:08 ;-) Feb 15 13:03:00 if someone finds a more recent version of the sdk for bcm47xx than 5.100.138.9, I am relay interested in it Feb 15 13:32:54 KanjiMonster: would openwrt work if I generate the images in the same order as brcm ref ? Feb 15 13:33:41 merbanan: no, the flash driver/partition parser doesn't handle this currently and it will fail to find the rootfs Feb 15 13:54:01 Hauke, [florian]: I'm back again. Would you please look at the bootlog of my dir-330 ( http://pastebin.com/Bv23E3KL ) to point me to the right direction Feb 15 13:54:15 I really need this box to work Feb 15 14:17:59 juhosg * r30549 /trunk/target/linux/ramips/files/drivers/net/ramips.c: ramips: raeth: serialize ramips_link_adjust calls Feb 15 14:18:02 juhosg * r30550 /trunk/target/linux/ramips/files/drivers/net/ramips.c: ramips: raeth: force link down in ramips_phy_stop Feb 15 14:47:52 [florian]: ayt? Feb 15 15:53:26 Barvinok: could you past the output of "nvram show" to pastebin Feb 15 15:53:41 sure, one sec Feb 15 15:54:54 here it is http://pastebin.com/stH1HfEF Feb 15 15:55:56 from the looks of it I suspect dd-wrt left a bunch of trash there... Feb 15 15:57:12 I found the values I am looking for Feb 15 15:59:35 Hauke: what parts of a broadcom sdk are you interested in? Feb 15 15:59:59 i don't have one right now, just in case i get access to one ;) Feb 15 16:00:01 nbd: an et driver under gpl license Feb 15 16:00:10 gmac? Feb 15 16:00:13 yes Feb 15 16:00:29 ok. i've looked through some SDKs, but could only find proprietary variants of it Feb 15 16:00:52 i'll see if i can get access to some newer stuff Feb 15 16:02:05 Hauke, I can give you console access to this router over ssh, if that would help Feb 15 16:02:16 Hauke: in 5.110.27.9 it's still proprietary Feb 15 16:02:31 broadcom said that they are working on changing the license for et and other stuff, but I do not know how they want to distribute it or when so I think it will be in the sdk at some time Feb 15 16:04:20 nbd: is it somewhere freely available in the internal or do you have some internal access to it? Feb 15 16:04:30 internal access Feb 15 18:33:08 nbd * r30551 /trunk/package/mac80211/patches/300-pending_work.patch: ath9k: prevent writes to const data on AR9160 Feb 15 18:34:21 nbd: if I have an ar9160, I probably want the above patch? Feb 15 18:34:44 yeah Feb 15 18:34:58 though i guess as long as there are no memory protections for const data, the bug is harmless Feb 15 18:35:14 unless you happen to have multiple ar9160 in one system Feb 15 18:35:20 ah no Feb 15 18:35:29 only 1, the other is ar9285 afaik Feb 15 18:36:41 then the bug probably doesn't do anything on your system Feb 15 18:37:04 ok so no need to hurry. thanks for the info Feb 15 18:44:34 build #115 of ar7 is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/ar7/builds/115 Feb 15 18:57:00 jow_laptop: is the new tweaked luci theme in svn yet? :) Feb 15 18:57:33 blogic: thanks for committing the ATM patches! Feb 15 18:57:46 philipp64|laptop: sure Feb 15 18:58:02 wondering if the Geode targets should bump to 3.2.5? Feb 15 19:01:12 it's .5 better, must be good! Feb 15 19:03:46 juhosg * r30552 /trunk/target/linux/ramips/files/drivers/net/ramips.c: ramips: raeth: use 're' instead of 'priv' everywhere Feb 15 19:03:48 juhosg * r30553 /trunk/target/linux/ramips/files/drivers/net/ramips_eth.h: Feb 15 19:03:48 ramips: raeth: add '__packed __aligned(4)' annotation to dma descritor structures Feb 15 19:03:48 This helps GCC to generare more efficient code. Feb 15 19:03:49 juhosg * r30554 /trunk/target/linux/ramips/files/drivers/net/ramips.c: ramips: raeth: use netdev_alloc_skb Feb 15 19:03:50 juhosg * r30555 /trunk/target/linux/ramips/files/drivers/net/ramips.c: ramips: raeth: add a helper function for skb allocation Feb 15 19:21:42 blogic: remind me... you own the builder list, right? Feb 15 19:22:41 not sure what you mean so probably not Feb 15 19:29:45 the regular builds of the tree... who does that? Feb 15 19:31:13 philipp64|laptop: yes, iirc Feb 15 19:31:43 philipp64|laptop: thepeople runs the buildbot (and is the only builder connected the last time I looked) Feb 15 19:33:07 ok. was wondering x86_alix2 could be added as a build target... will ask thepeople. Feb 15 19:35:37 * swalker would prefer if the non-building platforms were disabled temporarily until someone cares enough about fixing them Feb 15 19:39:06 yeah, that would seem to defeat the purpose otherwise.... how many don't build, anyway? Feb 15 19:41:26 it's rather skewed at the moment because of Werror Feb 15 19:43:20 swalker: i'll add -Wno-error=unused-but-set-variable to the cflags, that should fix most of the -Werror related nastiness Feb 15 19:43:43 dape_x60t: the bootstrap one? Feb 15 19:44:12 yep that one Feb 15 19:44:26 yes, committed it like one hour before Feb 15 19:44:36 I had to clean up a whole bunch of html to make it work properly Feb 15 19:44:44 erm s/before/ago/ Feb 15 19:44:58 ive read the list, anyways great job both of you, im gonna test it from tonight on Feb 15 19:49:21 nbd * r30556 /trunk/rules.mk: build: add -Wno-error=unused-but-set-variable to cflags for gcc 4.6 and newer to avoid stupid -Werror related package breakage Feb 15 19:49:38 there, that should make the buildbots a bit more happy Feb 15 19:49:53 nbd * r30557 /packages/net/batman/Makefile: batman: mark the gateway module as broken, it does not compile with any recent kernels Feb 15 19:49:56 nbd * r30558 /packages/utils/bluez-utils/Makefile: bluez-utils: add -lm to LDFLAGS to fix a compile error Feb 15 19:53:53 well x86 is fubar'd atm Feb 15 19:54:02 regarding non building platforms Feb 15 19:54:13 but stuff that could really go is ps3 or ubicom Feb 15 19:54:21 they didn't build in ages Feb 15 19:55:16 all x86? Feb 15 19:55:32 or just generic? Feb 15 19:55:41 probably all Feb 15 19:55:54 missing symbols Feb 15 19:57:32 build #114 of sibyte is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/114 Feb 15 20:40:39 build #138 of s3c24xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/138 Feb 15 20:43:12 jow_laptop: just built the alix2 target ... no issues. maybe something in my defconfig changed that... Feb 15 20:43:20 hmm, i wonder how i can make the luci bootstrap theme default.. Feb 15 20:43:51 maybe via files/etc/config/luci .. Feb 15 20:43:57 philipp64|laptop: well as previously stated, the issues become apparent if all packages are enabled, merely building a minimal set does not cause it Feb 15 20:44:36 ah. then I guess I don't understand the point of building each of the targets with all packages enabled... Feb 15 20:44:49 ? Feb 15 20:44:52 it only takes one such target to detect damage, right? more than that seems pointless. Feb 15 20:45:00 ?? Feb 15 20:45:20 it's deterministic. Feb 15 20:45:24 x86 has other modules Feb 15 20:45:30 that ar71xx or ramips Feb 15 20:45:39 it also has other symbols Feb 15 20:45:54 a mips-thingy won't ask for EFI support for example Feb 15 20:46:01 so I need to build with enable-all to test it? Feb 15 20:47:49 jow_laptop: if i may: enabling LED netdev support IN kernel_menuconfig conflicts with enabling in menuconfig at kernel modules? Feb 15 20:48:08 dape_x60t: try it Feb 15 20:48:11 philipp64|laptop: yeah Feb 15 20:48:32 i have, doesnt seem to do anything wrong, im just concerned about.. space.. Feb 15 20:56:23 well, I should have an answer back for Jo in about 16 hours when my build finishes... :-( Feb 15 20:57:36 I take that back. already finished. what else besides setting CONFIG_ALL=y do I need to do? Feb 15 20:58:53 CONFIG_ALL only has an effect if the .config is not populated yet Feb 15 20:59:16 I posted a sed oneliner here a while ago which simply enables any kmod-* in .config Feb 15 21:00:37 the general process is simple, install all packages from the packages feed to catch out-of-tree kmods as well Feb 15 21:00:46 run make defconfig to sync the .config Feb 15 21:01:09 then replace all # CONFIG_PACKAGE_kmod-* is not set with CONFIG_PACKAGE_kmod-*=m Feb 15 21:01:14 and start make Feb 15 21:01:34 you don't need to actually build everything, the missing symbols should already become apparent in the very early build stages Feb 15 21:04:42 so the missing symbols are all in linux then? Feb 15 21:05:20 yeah Feb 15 21:06:05 ok. Feb 15 21:06:14 for me target/linux/prepare is enough for triggering any missing symbol questions after selecting everything (and requires almost nothing, not even a built toolchain; just sed and some host tools) Feb 15 21:07:55 just make sure you call it with V=99 (somehow V=1 works as defconfig for the missing symbols, so you never get actively asked) Feb 15 21:08:30 nbd * r30559 /packages/sound/flite/Makefile: flite: add missing -lm Feb 15 21:08:33 nbd * r30560 /packages/net/fwknop/patches/001-fix_config.patch: fwknop: fix messed-up patch header Feb 15 21:12:24 say, on a multi-core, which PARALLEL symbols can I safely set? PKG_BUILD_PARALLEL? TOOLCHAIN_PARALLEL? Feb 15 21:12:57 both of these are much less tested Feb 15 21:13:11 compared to just using make -j Feb 15 21:14:29 ok. Feb 15 21:15:38 dape_x60t: space shouldn't be affected much. also recent openwrt trunk handles the case fine where packaged drivers are actually built into the linux image Feb 15 21:15:50 got it ! Feb 15 21:29:41 build #113 of xburst is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/113 Feb 15 21:43:20 florian * r30561 /trunk/target/linux/ramips/ (11 files in 7 dirs): Feb 15 21:43:20 [ramips] add Ralink V11ST-FE board support Feb 15 21:43:20 This board is equipped with: Feb 15 21:43:20 - 4MB of NOR Flash Feb 15 21:43:20 - 32MB of SDRAM Feb 15 21:43:21 - IC+ IP175C switch Feb 15 21:43:21 - one Mini-PCI slot Feb 15 21:43:29 florian * r30562 /trunk/target/linux/ramips/files/arch/mips/pci/pci-rt288x.c: [ramips] correct PCI IO/Mem resource swapping Feb 15 21:49:13 juhosg * r30563 /trunk/target/linux/ramips/files/arch/mips/pci/pci-rt288x.c: ramips: rt288x: fix PCI resource conflict Feb 15 22:12:38 jow_laptop: uhm, a little.. weird problem with bootstrap luci theme, i have right now a wireless linked android tablet to the wr703n running it and it seems i cant really choose a submenu from neither Status or System or the other tabs, as soon as i touch-click to open the submenus it just loads the current page and .. Feb 15 22:13:18 tried with the asus tablet in the keyboard dock with touchpad, same experience, as soon as i click with mouse pointer the tab to open submenus it just reloads and.. Feb 15 22:14:02 well I suppose touch doesn't work at all with it Feb 15 22:14:56 i'll fire up windows7 station to check it, i suppose you cant just add another submenu with the same name as the tab and make it so it wont refresh page on click ..? Feb 15 22:15:55 well its the same problem the old dropdown menu thing had with touch devices Feb 15 22:16:01 thats why I switched it to tabs Feb 15 22:16:18 the default openwrt theme works great, i can choose them each without problems Feb 15 22:16:30 yeah because it does not utilize hover events Feb 15 22:16:38 which do not apply to touch interfaces Feb 15 22:16:52 its np anyway, its a refreshing theme Feb 15 22:18:01 hmm Feb 15 22:18:07 damn, i forgot to update the romanian lang for a while Feb 15 22:18:09 the dropdown menus work fine on my iphone Feb 15 22:18:22 touching them does not cause a page reload Feb 15 22:18:29 uhm, my android default browser sends "tablet" for css Feb 15 22:18:38 shouldn't matter Feb 15 22:18:41 let me try "desktop" mode Feb 15 22:18:44 the css has no device specific handling Feb 15 22:18:48 hmm Feb 15 22:21:11 meh, iphone 3gs works like yours, dont know then, android browser... Feb 15 22:25:34 hey, guys! what's wrong with my patches for tayga? Feb 15 22:28:52 raorn: nothing, you may blame me for not looking into it Feb 15 22:29:04 ok Feb 15 22:29:17 * raorn blames jow_laptop fot looking into it Feb 15 22:29:20 ;-) Feb 15 22:38:50 * kerneis blames jow_laptop for not looking into tcpdump version bump :-) Feb 15 22:49:32 jow_laptop: building 3.2.2 stalls on the following: Feb 15 22:49:37 SCSI_MVSAS_TASKLET Feb 15 22:49:39 MD_LINEAR Feb 15 22:49:40 MD_RAID0 Feb 15 22:49:42 MD_RAID1 Feb 15 22:49:44 MD_RAID10 Feb 15 22:49:45 MD_RAID456 Feb 15 22:49:47 MD_MULTIPATH Feb 15 22:49:48 SKGE_GENESIS Feb 15 22:50:46 can someone add these to target/linux/x86/config-3.2 please? Feb 15 22:56:02 I'm actually not sure about the md stuff Feb 15 22:56:10 we have some md modules packaged Feb 15 23:01:19 philipp64|laptop: can you please try this? http://luci.subsignal.org/~jow/block-drop-m.patch Feb 15 23:02:18 philipp64|laptop: and rm -r tmp/ to rule out caching effects Feb 15 23:06:14 jow_laptop: rebuilding from scratch... Feb 15 23:22:57 i've been reporting the SCSI_MVSAS_TASKLET and SKGE_GENESIS on alix2 for a week or two now Feb 15 23:26:13 russell--: I know :) Feb 15 23:26:53 philipp64|laptop: config symbols that only pop up when selecting certain kmods (and aren't/shouldn't be handled by kmods) should go into generic/config-*, else they will get automatically removed again when doing kernel_*config Feb 15 23:37:03 hard to know which ones are specific to the x86... I figure if they were non-platform symbols, someone would have fixed them by now. Feb 15 23:38:24 i don't see them when building for x86-net45xx Feb 15 23:38:52 older kernel, i guess Feb 15 23:39:08 root@hurt:~# uname -a Feb 15 23:39:08 Linux hurt 2.6.39.4 #13 Tue Feb 14 23:31:06 PST 2012 i486 GNU/Linux Feb 15 23:41:11 There's nothing x86-specific about SCSI_MVSAS_TASKLET Feb 15 23:41:38 ah... but it does depend on PCI. Feb 15 23:48:16 russell--: see patchwork 1878. Feb 15 23:49:27 I thought about doing specific targets (instead of just profiles) for the 4801 and 4501, but since they've been out of production for so long--and since getting information from Soekris is such a PITA--I decided it wasn't worth the trouble. Feb 15 23:52:04 jow_laptop: at what point do I know if the build fails or not? do I need to let it build all of world? Feb 15 23:54:11 BATTERY_BQ20Z75, SENSORS_K10TEMP, SENSORS_PKGTEMP, SENSORS_VIA_CPUTEMP just showed up in a vanilla all-packages build of r30543 for net4521 Feb 15 23:57:21 philipp64|laptop: what information do you want? Feb 15 23:57:22 I thought I fixed those all already? Feb 15 23:57:34 EqUaTe: huh? Feb 15 23:57:39 though they won't do much more than ~10 megs imo... Feb 15 23:57:58 my work has a buttload of 4801s Feb 15 23:58:10 oh, for the 4801? Feb 15 23:58:33 Well, the switch for instance gets initialized by BIOS as a hard reset, but there is a way to tickle it to put it under software control. Feb 15 23:58:43 I've asked Christian how to do this, but never heard back. Feb 15 23:59:35 ah Feb 16 00:00:15 hmmm... http://www.fpaste.org/IIkW/ Feb 16 00:01:11 build #113 of ixp4xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/113 Feb 16 00:02:49 russell--: it looks like r29953 clobbered SENSORS_PKGTEMP ... Feb 16 00:02:51 philipp64|laptop: the log looks fine, apparently it passed all kconfig questions Feb 16 00:03:07 philipp64|laptop: did it still query MD stuff? Feb 16 00:03:13 right, but mac80211 seems to be broken... Feb 16 00:03:26 with your patch? it doesn't query for the MD stuff, no. Feb 16 00:03:49 but still SCSI_MVSAS_TASKLET and SKGE_GENESIS ? Feb 16 00:03:56 just mailed out a patch for the SKGE and SCSI_MVAS_TASKLET stuff... 1878... Feb 16 00:04:01 yes. Feb 16 00:04:04 ok Feb 16 00:04:36 jow * r30564 /trunk/package/kernel/modules/block.mk: [package] kernel: fix KCONFIG depends of kmod-md modules Feb 16 00:06:23 and... eglibc breaks. Feb 16 00:06:25 grrrrr. Feb 16 00:07:55 jow * r30565 /trunk/target/linux/generic/config-3.2: Feb 16 00:07:55 linux 3.2: missing generic symbols Feb 16 00:07:55 These symbols are dependent on CONFIG_PCI being enabled. Feb 16 00:07:55 Signed-off-by: Philip Prindeville Feb 16 01:13:13 jow_laptop: one broken module: Feb 16 01:13:28 ERROR: module '/home/philipp/openwrt-geos/build_dir/linux-x86_geos/batmand-r1439/batman/linux/modules/batgat.ko' is missing. Feb 16 01:13:30 make[3]: *** [/home/philipp/openwrt-geos/bin/x86/packages/kmod-batgat_3.2.2+r1439-2_x86.ipk] Error 1 Feb 16 01:13:31 make[3]: Leaving directory `/home/philipp/openwrt-geos/feeds/packages/net/batman' Feb 16 01:13:33 make[2]: *** [package/feeds/packages/batman/compile] Error 2 Feb 16 01:13:34 make[2]: *** Waiting for unfinished jobs.... Feb 16 01:22:08 nbd: update: 23 hours and counting 0 of 5 devices with your ath5k calibration fixes have failed (none of the 4 devices testing the first patch had failed at this point either though) **** ENDING LOGGING AT Thu Feb 16 02:59:57 2012