**** BEGIN LOGGING AT Mon Jun 06 02:59:57 2011 Jun 06 04:30:55 build #37 of ixp4xx is complete: Failure [failed shell_4] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/37 Jun 06 05:06:12 build #44 of ps3 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/44 Jun 06 05:46:25 build #44 of pxcab is complete: Failure [failed shell_4 compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/44 Jun 06 06:07:31 build #34 of iop32x is complete: Failure [failed shell_4] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/34 Jun 06 06:15:20 build #35 of au1000 is complete: Failure [failed shell_4] Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/35 Jun 06 07:09:54 build #41 of uml is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/41 Jun 06 07:28:14 so, how do i figure out where the knobs are for the second ar8316 switch chip on this routerboard 493g? there are 7 mdio devices or ports or whatever: x00, x01, x02, x03, x04, x1e, x1f. Jun 06 07:28:44 only the ar8316 associated with eth1 is found Jun 06 07:36:02 russell--: you need to probe the bus Jun 06 07:36:11 and if that fails it might be a gpio based mdio Jun 06 07:43:31 blogic: i'm looking at target/linux/ar71xx/files/arch/mips/ar71xx/mach-rb4xx.c Jun 06 07:46:04 what does probing the bus look like? Jun 06 07:48:47 the code there for the 450g seems to set a phy_mask and then call ar71xx_add_device_eth Jun 06 07:58:30 russell--: not sure how ar71xx does it Jun 06 07:59:23 russell--: https://dev.openwrt.org/browser/trunk/target/linux/lantiq/patches-2.6.39/0011-MIPS-Lantiq-Add-ethernet-driver.patch Jun 06 07:59:43 search for ltq_etop_mdio_probe Jun 06 08:00:02 390 -> 400 has an example of how i do it on lantiq/ based boards Jun 06 08:09:17 russell--: the ar8316 driver currently only attaches as switch to phy_id 0 Jun 06 08:11:16 KanjiMonster: /me is confused about the mdio ports, what do they represent? Jun 06 08:12:29 russell--: http://en.wikipedia.org/wiki/MDIO#Registers ;) Jun 06 08:12:54 or better http://en.wikipedia.org/wiki/Media_Independent_Interface Jun 06 08:13:24 russell--: at mdio address 00 to 04 are the phy registers of the five ports of the switch chip Jun 06 08:13:34 okay Jun 06 08:13:50 what are the 0x1e and 0x1f, do you think? Jun 06 08:14:46 good question Jun 06 08:15:39 mdio addresses 0x10 - 0x18 are used for accessing the switch registers Jun 06 08:16:41 russell--: you could try looking at the phy_id registers at 2 and 3; these are used to identify the phy device behind an mdio address (or should at least tell you the manufacturer) Jun 06 08:18:34 how do i look at them? is there a sysfs interface for that or do i start sprinkingly printk's? Jun 06 08:20:55 russell--: hm, I'm not sure - in worst case you can do some printks Jun 06 08:23:58 russell--: generally to you original question: https://dev.openwrt.org/browser/trunk/target/linux/ar71xx/files/arch/mips/ar71xx/mach-ubnt.c#L185 - the add_device_mdio set the /ignore/ mask when scanning the bus; bit(x) in the mask says ignore phy address x and don't scan it; while the phy_mask says the opposite and tells the driver at which phy address(es) its phy device(s) are Jun 06 08:24:43 ideally mdio_mask & phy_mask == 0 ;) Jun 06 08:25:45 there is no driver symlink in /sys/devices/platform/ag71xx-mdio/ag71xx-mdio:1e and :1f, while the others (00->04 are driver -> ../../../../bus/mdio_bus/drivers/Atheros AR8216!AR8316) Jun 06 08:25:50 okay Jun 06 08:26:09 so, would it make sense to play with those numbers? Jun 06 08:26:23 like, try different values? Jun 06 08:27:00 different values for phy_mask, that is Jun 06 08:29:10 russell--: sure; you could try if eth0 attaches anything to these two Jun 06 08:29:47 my starting place is a patch i found on the openwrt forums Jun 06 08:30:03 * russell-- searches through my many open tabs to find the link. ... Jun 06 08:31:05 https://forum.openwrt.org/viewtopic.php?id=28889 Jun 06 08:46:48 KanjiMonster: rb493g_setup uses BIT(0) for eth0, but the syslog messages say eth0 connected to PHY at ag71xx-mdio:04, and uses BIT(4) for eth1 but syslog says eth1 connected to PHY at ag71xx-mdio:00. Jun 06 08:50:41 * russell-- hopes that is not as baffling for others as it is for me Jun 06 09:33:56 hmm, okay, seems like fiddling with the phy_mask is the wrong direction Jun 06 09:34:31 * russell-- looks closer at target/linux/generic/files/drivers/net/phy/ar8216.c Jun 06 09:37:29 russell--: ar8216's config_init checks if the phy_address is 0 - if yes, then it registers the switch; else it behaves like a standard phy driver (i.e. provides link status etc, but nothing fancy) Jun 06 09:47:57 KanjiMonster: pdev->addr is phy_address? Jun 06 09:48:13 right Jun 06 09:48:16 gotcha Jun 06 09:53:39 KanjiMonster: going to try to just bypass that if (pdev->addr != 0) section, see what happens Jun 06 10:11:11 KanjiMonster: that does indeed register both switches Jun 06 10:12:21 not seeing a carrier on eth0 though Jun 06 10:12:35 despite a cable being plugged Jun 06 10:38:24 KanjiMonster: do you think the TEE can be de-ipv6-ified in the same way you did for bridge.ko? Jun 06 10:38:41 though I'm not sure whether its really worth it Jun 06 10:38:51 thats like shipping tcpdump which can only see v4 traffic Jun 06 10:45:56 xMff: I think with TEE it would be cleaner to split it into two (three) modules; the IPv4 and IPv6 are already separate xt_targets, so it shouldn't be hard to break out ipv6 TEE into its own module (although the xt_TEE looks like it was like this previously and got merged, looking at the module_aliases) Jun 06 10:46:47 problem is that it would pretty much violate all documentation out there Jun 06 10:47:00 TEE vs. TEE4 and TEE6 Jun 06 10:49:23 that would only affect the kmod names Jun 06 10:49:52 see http://lxr.linux.no/#linux+v2.6.39/net/netfilter/xt_TEE.c#L268 Jun 06 10:50:30 ok I see Jun 06 10:50:39 I will look into splitting it again then Jun 06 10:50:51 kmod-ipt-tee and kmod-ipt-tee6 sounds reasonable to me Jun 06 10:50:57 yeah Jun 06 10:51:17 or maybe kmod-ip6t-tee Jun 06 10:51:26 there are probably going to be more v6 specific ones Jun 06 10:56:20 xMff: you should find everything having the same issues by grepping for IPV6=n || IPV6 in Kconfig Jun 06 11:36:31 KanjiMonster: sorry had to go to work, yes will need to check Jun 06 15:14:04 <_lore_> ping nbd Jun 06 15:14:16 pong Jun 06 15:14:24 <_lore_> hi Jun 06 15:15:02 <_lore_> I tried the solution we agreed last week Jun 06 15:15:13 <_lore_> do you rember? Jun 06 15:16:33 <_lore_> but unfortunately it does not fix my issue Jun 06 15:16:56 <_lore_> I have another idea.. Jun 06 15:18:02 <_lore_> my feeling is that sometimes the HW is not yet finished to process the token frame in ath9k_rx_tasklet Jun 06 15:18:09 build #44 of lantiq is complete: Failure [failed shell_4 compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/44 Jun 06 15:18:52 <_lore_> and the frame processinf stucks because the device will not recive any other rx interrupt aafter this event Jun 06 15:19:11 <_lore_> is it reasonable? Jun 06 15:31:45 build #45 of cobalt is complete: Failure [failed shell_4] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/45 Jun 06 17:48:28 _lore_: interesting Jun 06 17:49:55 _lore_: could be Jun 06 19:30:06 hi, could varnish be interesting for anybody here? Jun 06 19:30:22 varnish ? Jun 06 19:30:28 as in the paint stuff ? Jun 06 19:30:45 http://www.varnish-cache.org/ Jun 06 19:30:55 a proxy cache Jun 06 19:31:16 slashdot uses it Jun 06 19:31:17 <_trine> I have a few old arm chairs :) Jun 06 19:32:29 <_trine> will there ever be a polyurethane.cache Jun 06 19:33:50 ok Jun 06 19:42:51 heilterminalcons: well, we have no policy to prevent it from being in packages/ feed Jun 06 19:43:14 so put it on the -devel mailing list and it will get merged if it meets the expectancies. Jun 06 19:45:36 okay thanks Jun 06 19:49:29 build #41 of ppc44x is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/41 Jun 06 21:01:41 u8 *eeprom = (u8 *) KSEG1ADDR(0x1fff1000); (for ar71xx platform) Jun 06 21:01:46 how is that address calculated? Jun 06 22:03:59 build #42 of x86 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/42 Jun 06 22:27:18 mirko * r27117 /trunk/package/pptp/files/options.pptp: [package/pptp] require mppe 128bit-encryption by default Jun 06 22:56:14 build #44 of orion is complete: Failure [failed compile_7] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/44 Jun 07 00:14:12 mirko * r27118 /trunk/toolchain/gcc/ (10 files in 2 dirs): [toolchain/gcc] upgrade Linaro GCC to 4.5-2011.05-0 - thanks to Mark Mentovai Jun 07 01:12:44 is the fact that I'm running on a rb493g accessible in the ar8216.c context? Jun 07 01:42:05 swalker * r27119 /packages/net/bwping/ (Makefile patches/): [packages] bwping: update to 1.3, drop obsolete patch, use PKG_INSTALL Jun 07 01:42:59 swalker * r27120 /packages/net/mercurial/ (Makefile patches/001-no_bzip2.patch): [packages] mercurial: update to 1.8.4, refresh patches Jun 07 01:43:41 swalker * r27121 /packages/net/subversion/Makefile: [packages] subversion: update to 1.6.17 **** ENDING LOGGING AT Tue Jun 07 02:59:57 2011