**** BEGIN LOGGING AT Mon Feb 27 02:59:58 2012 Feb 27 03:50:29 jow_laptop: ppc40x, ppc44x, brcm63xx, brcm47xx, etrax & orion all failed after r30708 Feb 27 04:57:49 swalker * r30735 /packages/libs/libv4l/ (4 files in 2 dirs): [packages] libv4l: update to 0.8.6, add v4l-utils package Feb 27 04:59:34 swalker * r30736 /packages/net/baresip/ (3 files in 2 dirs): [packages] baresip: fix avformat & v4l bustage Feb 27 07:41:16 build #127 of rb532 is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/127 Feb 27 08:24:37 build #99 of mpc52xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/99 Feb 27 09:30:17 build #142 of s3c24xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/142 Feb 27 09:31:04 hi Feb 27 09:32:00 is it possible that 16MB of RAM is not enough to run attitude adjustment? Feb 27 09:32:11 nettle: depends Feb 27 09:32:27 did you install lots of extra apps ? Feb 27 09:32:35 no nothing extra Feb 27 09:32:59 not even luci Feb 27 09:33:12 what arch ? Feb 27 09:33:20 ar7 Feb 27 09:35:20 [florian]: ping Feb 27 09:35:36 [florian]: have you tried AA on ar7 ? Feb 27 09:36:03 <[florian]> blogic: not yet Feb 27 09:36:25 <[florian]> but 16MB is usual on ar7 and never caused any problem Feb 27 09:37:37 nettle: do you have a bootlog ? Feb 27 09:37:53 <[florian]> what are the symptops exactly? out of memory? Feb 27 09:38:10 well i have bootlog Feb 27 09:38:16 <[florian]> please show it to us Feb 27 09:39:41 i even created a ticket but i was log time ago and nothing have changed since Feb 27 09:39:42 https://dev.openwrt.org/ticket/9565 Feb 27 09:40:39 <[florian]> the issue is with the switch Feb 27 09:40:53 <[florian]> and yes it's still present, and yes I still have not fixed this properly Feb 27 09:41:12 ah ok tx Feb 27 09:41:54 can i somehow disable it to get it working? Feb 27 09:42:03 <[florian]> no Feb 27 09:42:14 <[florian]> you'd certainly have to apply some patches that used to be in trunk Feb 27 09:42:30 <[florian]> what kind of switch do you have? Feb 27 09:42:38 <[florian]> some marvell 88e6060? or something different? Feb 27 09:42:58 ADM6996M Feb 27 09:43:14 <[florian]> ok, that's why it does not work Feb 27 09:43:52 <[florian]> nettle: try to remove the 972 and 973 pathces in target/linux/ar7/patches-2.6.37/ Feb 27 09:44:20 oki, and tx Feb 27 09:44:38 <[florian]> you are welcome Feb 27 10:03:53 jow * r30737 /trunk/target/linux/generic/ (5 files in 2 dirs): [generic] partly revert r30708, it breaks various arches Feb 27 10:04:47 jow * r30738 /trunk/target/linux/generic/files/drivers/net/phy/rtl8366_smi.c: [generic] revert unrelated change in previous commit Feb 27 10:17:04 hmm, a very strange thing happening in the trunk from last 2-3 weeks, seems that pppoe-won wont get the /128 ipv6 address advertised from the ISP anymore Feb 27 10:17:08 i can set it manually Feb 27 10:18:38 hmm, dont have iproute nor radvdump installed but im gonna do them and debug things out Feb 27 10:19:08 might be a sysctl issue Feb 27 10:19:20 accept_ra is on Feb 27 10:19:28 forwarding should be off and accept_ra should be 1 or 2 Feb 27 10:19:40 what was 2 for. Feb 27 10:19:49 2 is to override forwarding Feb 27 10:20:13 wait, we dont want ipv6 forwarding on ipv6 ? Feb 27 10:20:20 not on your wan Feb 27 10:21:29 so forwarding should be 0 on pppoe-wan Feb 27 10:21:39 yes Feb 27 10:21:58 it is 1 right now, as accept_ra is.. how can i set 0 for pppoe-wan ? i mean, what if pppoe-wan is down .. Feb 27 10:22:10 should i just edit /etc/sysctl.conf with pppoe-wan there? will it work? Feb 27 10:22:31 uci set network.wan.ipv6_accept_ra=1; uci set network.wan.ipv6_forwarding=0 Feb 27 10:22:42 ah, using generic wan name, thanks! Feb 27 10:22:47 * dape-work slaps himself Feb 27 10:23:08 hmm, wait, why aint these by default in /etc/sysctl.conf ? Feb 27 10:23:35 build #141 of at91 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/141 Feb 27 10:25:20 hmm, nice uci saved them in /etc/config/network under wan interface Feb 27 10:25:48 build #139 of ubicom32 is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/139 Feb 27 10:33:53 hmm, so i have option ipv6_accept_ra '1' and option ipv6_forwarding '0' under wan section for my pppoe-wan in /etc/config/network but the ipv6 address aint there and i still can see 1 in both /proc/sys/net/ipv6/conf/pppoe-wan/forwarding .. Feb 27 10:33:59 did uci commit, reboot.. Feb 27 10:34:33 im using netifd if that matters.. Feb 27 10:35:46 netifd does not handle those I believe Feb 27 10:36:12 try to set the ctls from within /etc/ppp/ip-up Feb 27 10:36:24 see if it helps Feb 27 10:44:28 ok, so /etc/ppp/ip-up not /etc/ppp/ipv6-up right? Feb 27 10:44:51 i know you mean what you say jow_laptop , just checking :P Feb 27 10:44:52 both would work I believe Feb 27 10:54:46 meh, put echo 0 > /proc/sys/net/ipv6/conf/pppoe-wan/forwarding in /etc/ppp/ip-up , restarted wan and forwarding still 1 :) Feb 27 11:02:33 meh, after ifup wan forwarding gets 1 Feb 27 11:04:53 <_trine> is mazilo still looking after FreeSwitch in openwrt? Feb 27 11:13:25 geez, cant even use tcpdump on pppoe-wan Feb 27 11:13:48 "g is not supported for link type LINUX_SLL: use -w" what the heck.. Feb 27 11:32:41 hmm, jow_laptop might be a firewall issue, i stopped firewall and ipv6 address was autoconfigured on pppoe-wan Feb 27 11:36:33 nbd * r30739 /branches/backfire/target/linux/generic-2.6/files/drivers/net/phy/ (rtl8366rb.c rtl8366s.c): Feb 27 11:36:33 generic: rtl8366{s,rb}: remove the PHY driver. (backport of r26600) Feb 27 11:36:33 Since the PHY driver is only used for the WAN port and there is virtually Feb 27 11:36:33 no difference between it and the generic PHY driver, we can sefely remove Feb 27 11:36:33 it. Feb 27 11:36:34 Signed-off-by: Jonas Gorski Feb 27 11:42:48 jow_laptop no icmpv6 type 10 for router solicitations on wan .. ? Feb 27 11:48:00 nbd * r30740 /branches/backfire/package/mac80211/patches/ (2 files): ath5k: merge calibration fixes from r30623, r30624 Feb 27 11:50:23 dape-work: i'll fix up the missing sysctl options later Feb 27 11:50:51 nbd uhm, do you know exactly where's the problem? cause i think i messed up stuff here, gotta try again and debug Feb 27 11:51:17 no idea about the other issues Feb 27 11:51:43 i'm just talking about the things that you appear to have worked around already Feb 27 11:53:24 uhm, stopped firewall a bit earlier and got the ipv6 address on pppoe-wan, but i think i messed up setting in /etc/config/network meanwhile cause with firewall down im not getting the ip6 anymore Feb 27 11:53:45 option accept_ra '1' option ipv6_accept_ra '1' and option ipv6_forwarding '0' Feb 27 11:56:08 meh, right, had router_solicitions set to 3 :) i tried 1000 and it got me the ipv6 address on pppoe-wan Feb 27 11:56:26 so i guess i will wait for firewall modification then, thanks nbd Feb 27 11:58:17 hmm, but i see router-solicitation is defined in /etc/config/firewall ... Feb 27 12:00:23 ah got it, should be on output WAN not input :) Feb 27 12:01:55 no, it says option src wan,gaaahh, im dizzy, gonna leave soon and go to sleep then Feb 27 12:04:23 list icmp_type router-solicitation Feb 27 12:09:34 yep, saw it, goddamit, i cant reproduce anymore Feb 27 12:10:03 btw im tcpdumping my ppp interface on provider side (i have access there) and i see RA sent regularly Feb 27 12:10:51 if you can't reproduce it now anymore it might be because relevant packets are conntracked or something Feb 27 12:10:57 try to reproduce after a router reboot Feb 27 12:33:32 ok, doesnt work yet, but, watching the RA sequence send by the ISP i can see on my firewall with ip6tables -L zone_wan -vx that the last get incremented Feb 27 12:33:40 input_wan and zone_wan_REJECT Feb 27 12:34:43 input_wan seems clear but zone_wan_REJECT has 7 616 reject all pppoe-wan any anywhere anywhere Feb 27 12:35:27 on the ISP side i can see whats rejected: 14:36:59.633838 IP6 (hlim 255, next-header: ICMPv6 (58), length: 48) fe80::1 > ff02::1: ICMP6, router advertisement, length 48 Feb 27 12:35:27 hop limit 64, Flags [none], pref medium, router lifetime 30528s, reachable time 0s, retrans time 0s[ndp opt] Feb 27 12:36:37 it takes a while until pppoe-wan is joined into the firewall Feb 27 12:36:50 so might well be that one or two RAs get dropped Feb 27 12:37:21 however I believe looking for RAs is a red herring; IPv6 on PPPoE happens via IPCP6, not ICMPv6 Feb 27 12:38:46 ipcp6 goes to ipv4 firewall ? Feb 27 12:39:11 its embedded in the pppoe session Feb 27 12:39:24 so it shouldn't get firewalled at all Feb 27 12:39:38 uhm, so far all i can see is that each RA from ISP goes to wan reject.. Feb 27 12:39:58 yes but as I said, IPv6 on PPPoE does not use RAs Feb 27 12:42:51 im really confused then, seems zebra sends RAs over this ISP ppp interface Feb 27 12:43:41 so are we talking about outgoing or incoming RAs? Feb 27 12:45:07 on the isp side zebra sends RAs over this ppp link, on client side i should autoconfigure the ipv6 ip after pppoe is up, in my particular case this openwrt up-to-date trunk wr740 should autonconfigure pppoe-wan after auth, right? Feb 27 12:45:50 i think incoming RAs to my openwrt router, yes Feb 27 12:46:12 am i seeing right that they are not allowed in the firewall? Feb 27 12:46:27 according to the rules, incoming RAs are allowed Feb 27 12:50:45 zone_wan chain, right? i dont see any router-advertisement icmp type.. Feb 27 12:56:24 yep, it gets firewalled, let me try a custom accept rule Feb 27 13:04:52 added a rule for input and the router-adv icmp type, RAs get matched there but still wont autoconfigure pppoe-wan, still debugging Feb 27 13:47:41 nah, its all messed up, i tried ra_accept 2 then increased router_solicitations to 1000 Feb 27 13:47:58 it eventually autoconfigures the ipv6 address on pppoe-wan but seems random Feb 27 13:56:56 kernel quirks prolly Feb 27 14:02:49 got it: two things changed, 1. router-advertisement icmp type allowed in firewall and 2. accept_ra 2 Feb 27 14:25:05 nbd maybe you can implement accept_ra 2 in netifd ? im looking at a base-files-network/files/etc/hotplug.d/iface/10-routes which has a small code for accept_ra 2 Feb 27 14:25:15 i will Feb 27 14:25:21 and if its 2 then it does do_sysctl "net.ipv6.conf.$DEVICE.accept_ra" $accept_ra Feb 27 14:25:25 aight, thanks! Feb 27 16:53:56 nbd * r30741 /trunk/package/netifd/files/usr/ (. share/ share/udhcpc/ share/udhcpc/default.script): netifd: add a new udhcpc default script for command line use (no uci support) Feb 27 16:54:00 nbd * r30742 /trunk/package/mac80211/ (36 files in 2 dirs): mac80211: update to wireless-testing 2012-02-23 Feb 27 16:54:04 nbd * r30743 /trunk/package/netifd/files/etc/hotplug.d/iface/10-sysctl: netifd: add a hotplug handler for setting sysctls (based on the old 10-routes file) Feb 27 16:54:07 nbd * r30744 /trunk/package/netifd/Makefile: netifd: update to 2012-02-26, fixes ipv4 broadcast handling Feb 27 16:54:18 dape: please test, the sysctl stuff should work now Feb 27 16:54:39 i simply copied the old code and ripped out the parts that are no longer relevant Feb 27 17:00:24 thank you, i will make an image for wr1043 right now Feb 27 17:08:01 grr Feb 27 17:08:34 no bovine sounds today? Feb 27 17:08:42 moo! Feb 27 17:08:46 eglibc build broken Feb 27 17:09:10 dl-profile.c:(.text+0x1a0): undefined reference to `__profile_frequency' Feb 27 17:09:10 dl-profile.c:(.text+0x668): undefined reference to `__profil' Feb 27 17:09:29 last time i built it, it still worked Feb 27 17:09:34 yes Feb 27 17:09:39 might need a dir clean Feb 27 17:09:56 I may have bumped binutils Feb 27 17:12:16 on a different topic, back to finding flash memory Feb 27 17:12:44 there's a struct that's visible via kallsyms on the vendor running kernel Feb 27 17:12:57 do you think I could put a module, force it to load, and get the values from it? Feb 27 17:13:08 this struct contains the pin mappings for the flash Feb 27 17:13:22 I've tried poking through the Image file without success Feb 27 17:14:06 only problem is that the data is in an __initdata section Feb 27 17:15:31 just note the address, decompress the kernel, hexdump it and jump to the offset Feb 27 17:15:58 tried, no joy Feb 27 17:16:04 ? Feb 27 17:16:10 there's code at that address, not data Feb 27 17:16:20 I've spend years peering at ARM code, so I know ;-) Feb 27 17:16:40 did you take the load address offset into account? Feb 27 17:17:02 I don't know Feb 27 17:17:05 will double check Feb 27 17:17:06 on many devices the kernel isn't loaded into physical ram offset 0 Feb 27 17:17:12 yeah, I know Feb 27 17:17:36 it's at c0008000 I think, common on ARM Feb 27 17:17:53 is it a uimage? Feb 27 17:18:02 or plain binary? Feb 27 17:18:07 I pulled the Image out of the uImage Feb 27 17:18:27 so uncompressed binary only Feb 27 17:18:27 got a link? Feb 27 17:18:36 jow_laptop: ping... got a question about dnsmasq configuration. Feb 27 17:18:52 the offset should be visible in the uimage header Feb 27 17:19:35 I don't have the Image etc or indeed the machine to hand. I might be able to get you an uImage. hang on. Feb 27 17:22:09 oh, I do have them. one minute. Feb 27 17:25:31 http://www.embeddedlinuxsolutions.com/bulogics/ Feb 27 17:25:37 I'm after the values at ek_nand_data Feb 27 17:26:13 they'll be, I think, 2 word-sized values somewhere in the 1-96 range Feb 27 17:26:38 sorry, 4 values. the first two may be 21, 22. Feb 27 17:47:29 nbd: ping Feb 27 17:47:33 Chocks: pong Feb 27 17:47:53 what was the kallsyms address of ek_nand_data? Feb 27 17:47:54 nbd: any comments on the files? Feb 27 17:48:00 during runtime Feb 27 17:48:11 hand on Feb 27 17:48:26 oh, I don't have that to hand, sorry. not until this evening Feb 27 17:48:47 ok Feb 27 17:48:55 i'll take another look when you have that Feb 27 17:48:57 just ping me again then Feb 27 17:49:02 very good Feb 27 17:49:04 nbd: any idea why Jo ended up quoting the section name on 'config' lines after we talked about only quoting values? Feb 27 17:50:01 my understanding was that only the value of 'option' and 'list' needed quoting. Feb 27 17:50:05 nbd: 3.2.5 kernel didn't boot at all, so need some work there somewhere Feb 27 17:52:28 philipp64|laptop: no idea, ask him Feb 27 17:52:50 did... didn't hear back. Feb 27 17:53:24 on another note, anyone have an idea why Adam is seeing tracebacks on uninstrumented builds, but when he turns on symbols they go away? Feb 27 17:53:49 philipp64|laptop: I already wrote that I did that to keep symmetry with options Feb 27 17:54:11 and to make it clear which fields are user content (within quotes) and which are structure (unquoted) Feb 27 17:54:31 oh, that's right. sorry, was doing something else at the same time we had that conversation (packing for a trip). Feb 27 17:54:32 I also said I don't agree with your statement of using an "inconsisten quote style" Feb 27 17:54:47 I'm not sure options need to be symmetrical with anything else. Feb 27 17:55:01 I am Feb 27 17:55:23 after spending almost two years with openwrt user support now I am fairly sure it is good the way it is Feb 27 17:55:39 values contain arbitrary strings, right? which need to be protected. and names are very limited in what they contain, and don't need to be protected. Feb 27 17:55:47 am I understanding that correctly? Feb 27 17:56:08 just because section names are cosntrained now doesn't mean they must be in the future Feb 27 17:56:36 also values don't need to be protected either Feb 27 17:56:41 only if they contain spaces Feb 27 17:56:53 however my opionion is set, I have no intention of changing it Feb 27 17:57:01 opinion Feb 27 17:57:02 not just spaces... wildcards, redirects, dollars, Feb 27 17:57:03 brb Feb 27 17:59:02 I was also thinking that the "config host / option ip / option mac / option name" followed by "config domain / option name / option ip" was a bit repetitive. Feb 27 17:59:33 why not add an option to "config host" like "option autopublish on" that implies the "config domain" part? Feb 27 18:00:13 or have a global option that says do name service for all mac's with reserved ip/name pairs? Feb 27 18:00:58 jow_laptop: so will it be possible to have 'config my$section' in the future? Feb 27 18:01:46 most current constraints are due to shell history, not an inherent format limitation Feb 27 18:03:58 right, which is why I mentioned dollars and redirects. Feb 27 18:04:49 I don't know why uci_escape() doesn't generate quotes when they are needed, and leave them out otherwise. Feb 27 18:05:21 because it would be effort Feb 27 18:05:24 then you could do other things, like have it handle base64 encoding (if you wanted) for arbitrary string values like PKI certificates, etc. Feb 27 18:05:53 you just said that you were including quotes now because they might be needed at some time in the future. Feb 27 18:06:13 seems a little ironic, if you're trying to avoid extra effort. Feb 27 18:06:16 :-) Feb 27 18:06:22 I also said that quotes now denote all palces which are user values Feb 27 18:06:37 I don't even know what that means. Feb 27 18:06:45 what's a "user value"? Feb 27 18:06:53 a value that isn't a keyword Feb 27 18:06:58 what if I add my own package that supports UCI in the future? Feb 27 18:07:04 or something that specifies structure Feb 27 18:07:11 no idea Feb 27 18:07:47 so it's not that easy to anticipate all scenarios. maybe it's just easier to deal with them when they come up. Feb 27 18:08:10 also having a printf("'%s'") is for sure less effort than if (quotes:_might_be_need) printf("'%s'", foo) else printf("%s", foo) Feb 27 18:08:15 you need an additional string scan Feb 27 18:08:17 if I submitted a patch that only added quotes when necessary (to user values), would you accept that? Feb 27 18:08:22 an additional format pattern Feb 27 18:08:40 depending on values the quoting style within an file is inconsistent Feb 27 18:08:57 no I wouldn't Feb 27 18:09:09 well, scanning a string for space, tab, parens, brackets, braces, dollars, pipes, etc. isn't that much work. it's 2 lines. Feb 27 18:09:38 I don't see how that's inconsistent. we have that issue in shell scripts. Feb 27 18:10:00 I don't follow your reasoning Feb 27 18:10:35 in other shell-based config files and scripts, most people know do do: Feb 27 18:10:43 this is not about generating the shortest possible representation of the data Feb 27 18:10:52 this is also not about making files "nice to read" Feb 27 18:10:52 export var=value var2="other value" Feb 27 18:11:19 and this discussion is getting tiresome Feb 27 18:11:30 you're optimizing stuff on the wrong end Feb 27 18:11:53 not sure what you meant by "after spending almost two years with openwrt user support now I am fairly sure it is good the way it is" if not that it's about 'nice to read'... or at least less error prone. Feb 27 18:12:32 people tend to copy & paste Feb 27 18:12:46 I'm trying to optimize it so that the output of UCI after parsing a script looks like the input. Feb 27 18:12:53 because it's less confusing. Feb 27 18:12:53 if they copy an unquoted line because thats all which is present in the file and then add a value with spaces it will be broken Feb 27 18:13:16 you add pointless micro optimizations Feb 27 18:13:25 comments are still stirpped Feb 27 18:13:29 order might still change Feb 27 18:13:30 and I'm assuming that the files/etc/config/* contents in the source tree are representative of what people write and what people expect to look at. Feb 27 18:13:33 whitespace still changes Feb 27 18:13:54 you're assuming wrong Feb 27 18:14:52 ok, so the configs in the source tree are arbitrary then. Feb 27 18:14:57 exactly Feb 27 18:15:18 and that's not confusing to users? Feb 27 18:15:33 which we also discussed. it has been pointed out that we should rather reformat the source files instead of changing the parser until it looks like human written files Feb 27 18:18:42 if I do: "touch file ; touch 'other file'" and then do "ls o" shell will expand to: Feb 27 18:18:52 ls "other file" Feb 27 18:19:24 right? but if I type: "ls fi" it expands to: Feb 27 18:19:25 ls file Feb 27 18:19:45 mine expands to ls other\ file Feb 27 18:19:48 because it drops the quotes where they aren't needed. I don't think anyone finds that confusing or ugly. Feb 27 18:19:51 don't see whats the point though Feb 27 18:20:06 it depends on whether you're using bash or tcsh, I guess. I use tcsh. Feb 27 18:20:37 I had users explicitely stating tzhat it is confusing that uci can have various kinds of quotings around values Feb 27 18:20:43 so? Feb 27 18:21:17 the point is that there's precedent for shortening the format to what's most likely to be entered by a user. Feb 27 18:21:29 and yes, having both single and double quotes is confusing. Feb 27 18:21:34 no Feb 27 18:21:36 that's easily solved, though. Feb 27 18:21:40 having quotes or no quotes is confusing Feb 27 18:22:53 perhaps for users who've never used sh... I can't believe that it's confusing for users with command line experience. Feb 27 18:23:43 anyway, what about the option to have a shorthand for "config domain"? Feb 27 18:25:06 and can a "config domain" section have multiple "option name" lines? Feb 27 18:25:38 the first one would be the A/PTR pair, and the successive ones would just be "CNAME" equivalents. Feb 27 18:26:17 i.e. option name canonical \n option name alias1 \n option name alias2 \n ... Feb 27 18:29:34 you can already specify multiple names Feb 27 18:31:58 ok, so which one does the address canonicalize back to? Feb 27 18:32:31 foooo Feb 27 18:33:10 the first Feb 27 18:34:32 ok. and auto-generating "config domain" sections? Feb 27 18:37:54 not sure what you mean Feb 27 18:38:11 jow_laptop: foooo ... you know ... the opposite of bar Feb 27 18:38:13 ;) Feb 27 18:40:39 see my message back at :59 ... Feb 27 18:41:36 if I already have "config host / option ip ... / option name ..." then it would be nice to not have to duplicate all of that with a "config domain / option ip ... / option name ..." with the same values. Feb 27 18:42:25 we could have an "option autopublish on" in "config dnsmasq" that controlled this. Feb 27 18:42:58 if this option was enabled, then every "config host" would have an implied "config domain" generated for it. Feb 27 18:43:00 why do you duplicate host sections into domain sections? Feb 27 18:43:34 because without them I'm not able to resolve the names of static dhcp hosts. Feb 27 18:44:14 I'd rather combine domain and host sections Feb 27 18:45:19 foo + bar ? Feb 27 18:45:28 or 'foo + bar' Feb 27 18:45:30 ok ok Feb 27 18:45:51 not sure that's wise. I also have domain sections for hosts that aren't dhcp served. Feb 27 18:46:22 so I can't deprecate the domain section altogether. Feb 27 18:46:50 sure. host sections without macs are domain sections Feb 27 18:46:59 host sections with macs are addtionally static leases Feb 27 18:47:09 host section with mac but no names are just static lease sections Feb 27 18:47:43 ok. that works. Feb 27 18:48:42 and ever think about generating a temporary config file for dnsmasq rather than passing everything on the command line? Feb 27 18:49:34 it makes it easier to start the server... copy aside the config file, make a change to the input config, reload the server, copy aside the 2nd config file, and diff the two together. Feb 27 18:49:36 yes, was too lazy to implement it Feb 27 18:49:52 also makes it easier to pastebin the resultant config. Feb 27 18:50:22 and less unlikely to overflow the command line limit on arguments to a process. Feb 27 19:00:10 jow_laptop: did you say a config domain block could have multiple 'option name' lines? I just tried that and only the last one works. Feb 27 19:01:47 it creates multiple -A /name/ip entries Feb 27 19:01:54 and a --ptr-record for the first one Feb 27 19:02:04 if that does not work its a dnsmasq issue Feb 27 19:02:13 as far as the config is involved it "works" Feb 27 19:03:15 ah, you need to do it via "list", not "option". Feb 27 19:03:55 the Wiki should probably explain that. Feb 27 19:04:20 option with space separated values works too Feb 27 19:04:46 and that's not confusing. :-) Feb 27 19:05:08 pardon? Feb 27 19:07:22 that an option field with spaces in it equates to a 'list' directive. Feb 27 19:07:47 thats a wrong conclusion Feb 27 19:07:59 depends on the config processor Feb 27 19:08:14 in this case its a for x in $(config_get ...) loop Feb 27 19:08:24 for which it is equivalent Feb 27 20:27:09 Feb 27 20:34:54 hi I need some help with JTAG/openocd. Im getting error "unexpected Feroceon EICE version signature" , not sure if I can ignore this or rather not... ? Feb 27 20:53:53 build #101 of adm5120 is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/101 Feb 27 21:44:52 build #96 of octeon is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/96 Feb 27 21:59:07 jow_laptop: Here (http://pastebin.com/uanB9mry) is the output of ./scripts/diffconfig.sh from my local OpenWRT root dir. Feb 27 22:32:09 does this code make any sense? Feb 27 22:32:12 http://pastebin.com/Y9TgqyQ4 Feb 27 22:33:35 it writes some bits to a register and reads 32bits from another Feb 27 22:43:54 it was my attempt to port this piece of code from the bootloader to the kernel Feb 27 22:43:58 http://pastebin.com/CqWBU2Em Feb 27 22:45:20 <[florian]> that ressembles it Feb 27 22:46:22 well the printk was diag_printf in the previous code Feb 27 22:46:30 <[florian]> it does not matter Feb 27 22:47:31 <[florian]> you are not getting the expected results? Feb 27 22:47:55 well, yes, and no Feb 27 22:48:17 with the current bootloader they doesn't match Feb 27 22:48:37 but with the original bootloader yes, the values matches Feb 27 22:48:40 <[florian]> have you thought about using CFE instead of RedBoot on your devices? Feb 27 22:49:11 yes, all CFEs bricked the board everytime I tried a CFE Feb 27 22:49:26 <[florian]> how come they brick your device? Feb 27 22:50:26 how? totally bricked Feb 27 22:50:38 <[florian]> that's not how, that's the result Feb 27 22:50:49 <[florian]> does it run once at least? Feb 27 22:51:22 yes, you're right, but I am only able to view the result Feb 27 22:52:18 even some bootloaders may load some code which makes difficult to restore the redboot again Feb 27 22:52:25 I mean at early startup Feb 27 22:53:19 maybe when the booloader initializes the RAM memory or who knows.. Feb 27 22:53:36 <[florian]> it certainly does Feb 27 22:54:50 uhm, guys , anyone notice anything strange with latest trunks? on the network chip ? i get eth down on wan port on wr740 and wr1043 with latest today trunk Feb 27 22:55:03 all on different locations, tried diff cables, plugged in diff switches Feb 27 22:55:22 could it be the RAM? I've noticed most CFE based boars use 143 MHz ram memory whereas the liveboxes use a 133 MHz one Feb 27 22:55:35 i can see eth1 on wr740n per example, but it is not up, i can ifconfig up it but as soon as i ifup wan it says eth1 down link not ready Feb 27 22:56:58 other difference is the physical start address of the flash chip, in the livebox is 0x1e400000 whereas other board is the well known 0x1fc00000 Feb 27 23:03:43 [23:45:35] <[florian]> does it run once at least? <-- no, never Feb 27 23:03:53 I forgot to answer Feb 27 23:23:51 nbd: ping please Feb 27 23:24:40 dape_x60t: pong Feb 27 23:25:18 thanks, well, if i try to ifup wan (configured as pppoe) it says eth1 link down after, if i configure static ip on eth1 i can arping via it.. Feb 27 23:26:13 its a wr740, latest trunk, after reboot eth1 goes down, i can ifconfig eth1 up and run ifup wan but right after it says eth1 link down Feb 27 23:26:17 nbd * r30745 /trunk/package/mac80211/patches/571-ath9k_beacon_fix.patch: ath9k: fix beacon issues on ar93xx Feb 27 23:26:20 nbd * r30746 /trunk/package/mac80211/ (21 files in 2 dirs): mac80211: update to wireless-testing 2012-02-27 Feb 27 23:26:28 hm, odd Feb 27 23:26:35 static seems fine on eth1 Feb 27 23:27:06 does it run pppd? Feb 27 23:27:29 sec,i rebooted it Feb 27 23:27:31 jow * r30747 /trunk/package/iwcap/ (. Makefile src/ src/iwcap.c): [package] add iwcap - a small radiotap capture utility for background monitoring of wireless traffic or use as a remote capture drone Feb 27 23:28:04 ok, eth1: link down after reboot Feb 27 23:28:29 still waiting for bridges to settle Feb 27 23:28:54 ok, done , i cant sett eth1 up, but its there at ifconfig Feb 27 23:29:21 show me the output of ifconfig eth1 Feb 27 23:30:23 http://paste.pocoo.org/show/557836/ Feb 27 23:30:38 that's ifup wan, not ifconfig eth1 Feb 27 23:30:55 and ifconfig eth1 http://paste.pocoo.org/show/557837/ Feb 27 23:31:29 changed cables, changed another media converter, tried it in a switch Feb 27 23:31:51 might be an issue in netifd Feb 27 23:31:55 need to check something Feb 27 23:32:07 sure, im here all night if you need me Feb 27 23:37:57 show me the output of logread Feb 27 23:38:13 all ? Feb 27 23:38:25 yeah Feb 27 23:38:31 jow * r30748 /trunk/package/iwcap/src/iwcap.c: [package] iwcap: fix reversed umask Feb 27 23:39:53 http://paste.pocoo.org/show/557839/ Feb 27 23:42:27 ok Feb 27 23:42:30 not a bug then Feb 27 23:42:38 Remote message: ^M^JAccess denied (external check failed).^J Feb 27 23:42:52 netifd detected that pppd reported authentication failed Feb 27 23:42:54 from pppoe, i connected the wan port to a switch not linked to my isp Feb 27 23:43:05 so it kept the interface down instead of repeatedly retrying Feb 27 23:43:10 hmm Feb 27 23:43:30 if i reboot it i will get eth1 down after thing settle Feb 27 23:44:03 yes Feb 27 23:44:05 that's normal Feb 27 23:44:11 netifd brings the device down when it's not in use Feb 27 23:44:22 aight, but pppoe credentials are ok Feb 27 23:44:50 let me try another account Feb 27 23:44:56 just to be 100% sure Feb 27 23:47:34 same thing, very strange, auth failed, http://paste.pocoo.org/show/557841/ Feb 27 23:48:52 the mac address seems to be corrupted for some reason Feb 27 23:49:12 trying without servicename and ipv6 Feb 27 23:49:25 ah, i just found a bug in netifd related to mac address handling Feb 27 23:49:48 neat Feb 27 23:50:17 wake up CIA-54 :P Feb 28 00:08:03 nbd * r30749 /trunk/package/netifd/files/lib/network/config.sh: netifd: add missing do_sysctl function Feb 28 00:08:06 nbd * r30750 /trunk/package/netifd/Makefile: netifd: update, fix some mac address handling issues Feb 28 00:08:22 shall i try? :) Feb 28 00:08:45 yes Feb 28 00:19:34 nbd: uhm, exactly the same behaviour Feb 28 00:20:02 show me the output of ifconfig -a Feb 28 00:20:18 sec please, rebooted to make sure Feb 28 00:22:55 nbd: http://paste.pocoo.org/show/557854/ Feb 28 00:23:33 eth1 doesnt go down anymore right after reboot but pppoe wont auth and go up and if i try ifup wan by hand.. Feb 28 00:23:34 ok, mac address corruption is gone Feb 28 00:23:48 please show me the output of logread again Feb 28 00:24:41 nbd: http://paste.pocoo.org/show/557855/ Feb 28 00:25:31 maybe the remote peer didn't like the mac address changes or something Feb 28 00:26:29 no mac protections Feb 28 00:30:25 maybe add some debug code to /lib/netifd/proto/ppp.sh to dump the pppd command that it's running Feb 28 00:30:38