**** BEGIN LOGGING AT Wed May 15 02:59:57 2019 May 15 09:14:56 KanjiMonster: i've checked 18.06 and that file does not exist in brcm63xx May 15 09:15:13 KanjiMonster: Also compile-testing this targets works fine May 15 09:15:27 I'll push as-is May 15 09:15:51 xback: I guess then that change was never backported. go ahead, no objections by me May 15 11:23:36 I'm getting some warnings about recursive dependencies with the latest master: https://pastebin.com/3CwjEx9C May 15 11:47:50 Has any one a target with mediatek mt7623? and also hitting the limit of 1gbit/s total? Even with multipul ports? May 15 11:59:21 Yes, I have a target with mt7623, the U7623 May 15 11:59:35 The switch is only 1GBit/s May 15 12:01:29 does it have the mt7530 with onit? May 15 12:01:54 Yes, it is equipped with mt7530 May 15 12:03:27 but link between soc and switch can be 2GBit/s May 15 12:04:19 Yes, but mt7530 is a gigabit switch May 15 12:07:20 Is the software also using the 2nd gmac for the wan port? May 15 12:09:27 I don't know, sorry :( May 15 12:10:28 No problem May 15 12:13:15 But I don't think so, at least not my device. I took a look at other attempts at working with the 2nd gmac on different devices using mt7530, and I didn't have to do anything special to make networking on my device work 100% May 15 12:14:17 I see that you were (probably) involved in some of the discussions/attempts I was refering to :) May 15 12:17:21 2nd gmac is working on my mt7621 board but also on bananapi R2 board with a mt7623 soc. May 15 12:18:58 But with mt7623 we are hitting 1gbit/s limit when using multiple iperf stream over multiple ports/gmacs. May 15 12:21:03 So I am wondering if someone else hit the same issue May 15 13:34:24 hi, what would be a Target System that uses the powerpc architecture? May 15 13:37:24 You can browse tmp/.targetinfo for one. apm821xx (AppliedMicro APM821xx) is an example. May 15 14:59:38 QCA9563 -- Data sheet that I found (Rev. D, April 2014) shows that it has some kind of hardware I2C (Section 8.2). I didn't find anything in the Linux DT bindings that looked applicable. Is a driver for that just under QCA SDK at this time? May 15 15:00:30 "the DW_APB_I2C interface" May 15 15:12:06 Grrr, "To access Synopsys protected applications, you must provide an Active Site ID in the field below." May 15 15:12:31 So much for that source of information May 15 15:14:57 which one of the iptable chains deals with packages forwarded from LAN to WAN? is it zone_lan_forward? May 15 15:17:15 Is there OpenWRT conventional way to make package Makefiles behave different depending on the kernel version? I tried ifdef based on LINUX_VERSION but it appears to fail. Presumably because that variable isn't defined everywhere. May 15 15:30:52 Neighbor11111115: you should be able to use $(CONFIG_LINUX_X_Y) e.g. ifeq ($(CONFIG_LINUX_4_14),y) \n do 4.14 stuff \n else \n do other kernel stuff \n endif May 15 15:31:19 charolastra: roughly speaking, yes May 15 15:33:29 thanks jow May 15 15:36:51 but i removed all rules in that chain and still see the responses from external servers when sniffing on the LAN interface May 15 15:41:27 responses from WAN will hit zone_wan_forward May 15 15:41:38 also, established connections skip all the zone_* chains May 15 15:41:50 or, by default, zone_wan_forward May 15 15:42:58 ok, but how block esteblished connections? May 15 15:43:17 delete the rule matching on RELATED,ESTABLISHED in the FORWARD chain May 15 15:43:47 sorry, FORWARD chain in the filter table May 15 15:47:24 will check, but in case of an empty zone_lan_forward, the packages are still forwarded, otherwise there wouldn't be a response from outside May 15 15:48:14 for new or established connections? May 15 15:49:22 new ones May 15 15:51:29 I tested on my own device now. If I flush zone_lan_forward and initiate a new connection, my router replies with Destination port unreachable May 15 15:57:07 hmm, after a reboot and flush i still get responses May 15 15:58:12 hm, that is indeed strange. Perhaps adding a log rule at the top of the FORWARD-chain will give some answers as to why packets are not dropped? May 15 16:08:58 good idea, i'll check tomorrow. thinking capacity for today is depleted May 15 16:59:58 shanee, it depends, since you would need to resolve the domains. If you know the addresses you want to allow/not allow, then you can skip the dnsmasq part May 15 17:00:03 sorry, wrong channel :) May 15 17:15:03 I just noticed that if a in-tree kernel module depends on a function defined by a kernel C file build as an object (instead of a .ko) then modpost will error with undefined if the .o file is not already built. However I thought OpenWRT build kernel modules before building the kernel so how did this ever work? Or am I misunderstanding something? thanks May 15 19:09:40 after 9 days DTAG finally seems to have fixed my inet May 15 19:09:53 been online for 12 hours with no disconnect May 15 19:10:28 all it took was 14 calls to the hotline, 3 technicians and an owrt dev to explain to them that the line card was bust May 15 19:11:15 the 3rd technician actually listened, let me explain to him how his diag box works then drove to the central exchange, repeated the test and then concluded *drumroll* that the line card was bust and replaced it May 15 19:12:30 yay telcos May 15 19:12:37 srsly May 15 19:12:43 the things the hotline told me May 15 19:13:12 best one was that they said it is normal that the inet gets a little slower everyday because the routers get older every day May 15 19:13:45 i had a support person suggest that the direction of a phone cord between the demarc and the dsl modem mattered May 15 19:13:55 or the technician that came to the house, touched the router, said its to hot, he can feel it and left telling me to buy a new one May 15 19:14:10 when they had really just turned off the service, against my persistent instructions May 15 19:14:14 russell--: and you should change the wireless cable May 15 19:14:25 or buy one of those ionizer sprays for the air May 15 19:15:32 anyhow, knock on wood,lets hope it lasts 5 more months when fiber finally arrives May 15 19:16:07 A while back I was unable to make connections to certain hosts over IPv6. Traceroute showed the packets were making it into the ISP network, so it was obviously their problem. I called support, and the first person told me my modem was broken and I needed a new modem. I knew that wasn't the case, so I kept pestering him until he let me talk to a higher-level technician, who agreed that my modem wasn't the cause and put in a ticket to get May 15 19:16:07 it fixed. Not ten minutes after I talked to him, it was working again. Three days later, I got a follow-up email saying they had investigated the problem and that my modem was broken and I needed a new modem… May 15 19:16:48 mamarley: it must have been too hot May 15 19:18:29 Guys, where I can request commit access to openwrt routing repository? Because I have there pull request since 14th January and I didn't get even respond to it. Why the repository isn't included under OpenWrt ( https://github.com/openwrt/ ) ? Also, it seems that routing packages are not actively maintained as community packages. Some of the routing packages were proposed to be included in community May 15 19:18:31 packages - https://github.com/openwrt/packages/pull/8855 see lynxis answer. Shouldn't we discuss it on mailing list? May 15 19:20:57 pepe it got merged 2 days ago ?! May 15 19:21:33 or do you have other PRs May 15 19:21:47 I'm not talking about smcroute, but about bird packages in routing repository May 15 19:37:23 blogic: it could be worse. After several weeks of failure to get a circuit running I pointed out to AT&T that they had the wrong mode fiber SFP transciever in their hardware so of course it couldn't establish a link. To fix the problem they had to cancel the circuit and provision a new one because their provisioning system did not allow changing the mode on an existing circuit. May 15 19:49:08 agb: nice one. May 15 20:16:42 Pepe: IMHO: the routing team should decide what they want to do. May 15 20:19:12 Yes, I agree on that, but it should be decided before you send pull request, where you removed smcroute from routing packages and then it was merged to packages May 15 20:22:26 Pepe: the maintainer of smcroute sad he won't maintain it anymore. Unmaintained packages should be removed or somebody else take the maintainership May 15 20:25:17 Pepe: somebody else (mwarning) later then created a PR to packages. May 15 20:59:17 Hm, with inactive maintainers you can remove many packages from community packages or routing packages. Yes, someone needs to take maintainership, but who? The guy without commit access? As I said earlier, there are 3 pull requests since 14th January and there is no response from the maintainer. Recently, what I have been able to find, you said in at least 3 posts that you would like to move all May 15 20:59:18 routing packages to community packages. Have you ever proposed it somewhere else than in comments? Would you please do it? However, I think it should be moved from OpenWRT-routing under openwrt and there should be routing packages, but I don't want to decide that. Currently, in routing packages, there are no checks, tests, nothing. May 15 22:04:08 Pepe: only checks are on the build bots **** ENDING LOGGING AT Thu May 16 03:00:30 2019