**** BEGIN LOGGING AT Wed Jun 13 03:00:02 2018 Jun 13 03:00:17 was the wifi overview page changed? i now have the problem that the overview shows the configured static dhcp lease instead of the actual client ip Jun 13 03:02:31 i have 2 networks with different subnets. on one network i have configured a static lease. when the client now joins the other network, dnsmasq will not give out the static lease, instead it gives out a ip from the pool. but the wifi overview page shows the configured static lease instead of the actual ip Jun 13 03:04:12 example: daemon.warn dnsmasq-dhcp[5915]: not giving name Dennis-Handy.lan to the DHCP lease of 10.0.1.101 because the name exists in /tmp/hosts/dhcp.cfg01411c with address 10.0.0.10 Jun 13 03:05:56 actual wifi overview https://ibb.co/gLjFaJ Jun 13 04:47:16 I am trying to change the mac addr of an interface: option macaddr but it's not changing. checking it via ifconfig Jun 13 05:07:38 I am building kernel with KPROBES. During build, kernel config goes in interfactive mode and asks for new options avalable like KPROBES_UVENT. I typically say y to them. But I have to do this every time I build the kernel. How can I take kernel's .config and use it as base config? Jun 13 05:16:26 I just copied .config to target/linux/armvirt/config-4.14 Jun 13 05:16:43 It took care of the problem but I don't know if thats the right way Jun 13 05:49:19 HSC: it is not as it will overrule a lot of settings that are usually merged from elsewhere Jun 13 05:50:15 HSC: the final config is usually composed from target/linux/generic/config-$version + target/linux/$target/config-* + target/linux/$target/$subtarget/config-* plus KCONFIG:=... symbols from KernelPackage Makefiles Jun 13 05:51:02 HSC: the best way is to use "make kernel_menuconfig" to activate kernel features, that will modify the kernel config template accordingly and you can review the results with "git diff" Jun 13 05:52:06 HSC: you can pass an additional variable "CONFIG_TARGET" that specifies which kernel configuration template to update Jun 13 05:53:37 jow: So in CONFIG_TARGET I can specify linux/armvirt/config-4.14? I thought it was to specify CPU architecture Jun 13 05:53:38 e.g. "make kernel_menuconfig CONFIG_TARGET=platform" will update target/linux/$target/config-* while CONFIG_TARGET=subtarget updates target/linux/$target/$subtarget/config-* Jun 13 05:54:00 nah, its just a badly named variable Jun 13 05:54:25 CONFIG_TARGET in this context means the target configuration file to update with "make kernel_menuconfig" Jun 13 05:54:41 jow: oh! Jun 13 05:54:42 not to be confused with the CONFIG_TARGET kconfig symbol :) Jun 13 05:55:03 I actually confused with that. Jun 13 05:55:08 understandable Jun 13 05:55:12 I think everyone is Jun 13 05:55:46 make kernel_menuconfig CONFIG_TARGET=armvirt is the correct command that I want then Jun 13 05:56:06 no, you need to pass a literal "platform" instead of "armvirt" Jun 13 05:56:46 "platform" selects target/linux/$(CONFIG_TARGET_BOARD)/config-* Jun 13 05:57:07 ok Jun 13 05:57:11 and "subtarget" selects target/linux/$(CONFIG_TARGET_BOARD)/$(CONFIG_TARGET_SUBTARGET)/config-* Jun 13 05:57:29 cool. I will try that Jun 13 05:57:32 thanks a lot! Jun 13 06:13:30 jow: I tried that. The armvirt/config-4.14 shows the change but it doens't show the new menu options thus while building kernel build again goes in interactive mode. Jun 13 06:14:41 I meant to say armvirt/config-4.14 shows *some* changes but no the new menuoptions Jun 13 06:15:26 While the genereated .config in kernel build directory has the new menu options Jun 13 06:32:08 HSC: ah, I think this is due to the fact that the openwrt buildsystem already handles kprobes Jun 13 06:32:57 HSC: in the normal menuconfig, Global build settings --> [ ] Compile the kernel with kprobes support Jun 13 06:33:17 so you do not need any customizations at all Jun 13 06:37:52 dedeckeh: is there any reason for actually building rdma for openwrt? afaict, the problem (which involves redundant struct definitions) only affects rdma, and rdma seems to be an infiniband thing, does any supported openwnrt devices have infiniband interfaces? Jun 13 06:46:10 ruseell--:to be honest no idea if rdma is used in the OpenWRT community. Maybe it's woth asking on the openwrt-devel mailing list Jun 13 06:46:19 *worth Jun 13 06:46:33 hmm i configured some arptable rules on my wan interface (docsis). in like 2 hours there are 770k dropped packets/22mb. is this normal for docsis? Jun 13 06:46:42 jow:ping Jun 13 06:54:41 hmm also why does mvebu use rx-usecs 100 and rx-frames 32, for tx there are all disabled. 100 seems also quite high= Jun 13 06:54:42 ? Jun 13 06:59:01 when I use option macaddr under wifi-iface, hostapd is crashing Jun 13 06:59:15 so... don't use it? :D Jun 13 07:00:20 jow: I am enabling anew knob CONFIG_KPROBE_EVENT which isn't there in openwrt kernel config Jun 13 07:00:28 thre is KERNEL_KPROBE_EVENT Jun 13 07:00:45 jwh: :) is there another way of changing wifi-iface macaddr? here is the log: https://gist.github.com/ironpillow/c3f4223eb6445b9d49bbe1437964b45d Jun 13 07:01:46 nope Jun 13 07:02:19 judging from the logs, probably not supported by driver/card Jun 13 07:03:11 jwh: ok. thanks! Jun 13 07:24:16 dedeckeh: pong Jun 13 07:24:33 russell--: I know one guy doing infiniband with openwrt Jun 13 07:25:49 HSC: then the cleanest solution is adding the extra new symbols to config/Config-kernel.in I'd say Jun 13 07:26:46 jow:Did you get feedback from netprince__ regarding the qos patch by quidosarducci ? Jun 13 07:27:05 HSC: CONFIG_KERNEL_KPROBE_EVENT in the openwrt .config is translated to CONFIG_KPROBE_EVENT in the kernel .config Jun 13 07:28:13 HSC: in fact any CONFIG_KERNEL_... is transformed to CONFIG_... in the kernel config Jun 13 07:28:21 dedeckeh: nope Jun 13 07:29:20 heh Jun 13 07:29:48 does this mean I have a error in my /etc/config/wireless file: daemon.err hostapd: Configuration file: /var/run/hostapd-phy0.conf Jun 13 07:30:26 ironpillow: not all wifi drivers support changing the mac address arbitrarily Jun 13 07:30:48 ironpillow: some have some very tight limits and a specific mask applied on top of the given mac address Jun 13 07:30:55 jow: i am not changing mac address anymore Jun 13 07:30:57 so you can only change the last two bits or so Jun 13 07:31:17 jow:PR has been opened fixing the qos issue https://github.com/openwrt/openwrt/pull/1044 Jun 13 07:31:22 the above line does not indicate any error Jun 13 07:31:29 daemon.err just means that the program logged to stderr Jun 13 07:38:34 jow: thank you :) Jun 13 09:19:30 jow: maybe the person doing infiniband should fix the include file problems ;-) Jun 13 09:34:53 ausjke: what's up with the lighttpd? Jun 13 10:21:28 the fix in master that dropped obsolete patches wasn't applied to 18.06, Jun 13 10:25:14 still hoping someone merges the device support for c7 v5, cpe210 v2 and 450e before a release...(well, 125 open requests.. but these 3 are my favorites) Jun 13 10:44:39 karlp: are you talking about lighttpd? Jun 13 10:44:51 yeah Jun 13 10:44:53 > ls net/lighttpd/patches/ Jun 13 10:44:54 0001-core-fix-base64-decode-when-char-is-unsigned-fixes-2.patch Jun 13 10:44:57 looks good to me Jun 13 10:45:19 it's from openwrt-18.06 branch Jun 13 10:46:09 * karlp shrugs, just repeating what ausjke was talking about yesterday and in their PR to try and reduce the loop turnaround time Jun 13 10:46:28 karlp: ok, thanks Jun 13 10:46:52 ausjke: i need you to clear it up for me, please :) Jun 13 10:47:11 do we have Hannu Nyman on IRC? Jun 13 10:48:03 oh, nevermind, I meant to complain about "Merge pull request #6281 from wvdakker/openwrt-18.06" Jun 13 10:48:13 rmilecki: nope, he's very active on the forum though and responsive when contacted via mail Jun 13 10:48:14 but I just noticed we have so mane shorewall packages Jun 13 10:48:15 ug Jun 13 10:48:23 *so many Jun 13 10:48:57 rotanid: blogic and me were planning a larger merge & review round next week Jun 13 10:49:03 until then, not much time Jun 13 10:49:13 heh, https://labs.mediatek.com/en/blog/using-mediatek-linkit-smart-7688-as-a-wi-fi-router-how-to-use-a-wi-fi-dongle suggests to just "keep typing 'wifi' until it comes up" Jun 13 10:49:31 somewhat unsurprisingly, that doesn't actually work very well / at all for me. Jun 13 10:49:48 is that rt2x00 ? Jun 13 10:49:55 rt2860 yeah, Jun 13 10:50:45 jow: ok, danke für die info Jun 13 10:51:05 should be the same as some of these huawwei echolife, which are brmc6368, with rt28xx wifi. Jun 13 10:51:18 wifi is detected, but doesn't ... work. Jun 13 10:52:33 possibly don't have the cal data loaded from flash maybe Jun 13 10:54:34 rmilecki: whats to complain about the shorewall merge? Jun 13 10:55:09 i thought there were 5 identical commits for version bump Jun 13 10:55:18 now I see these commits are for separated packages Jun 13 10:55:25 so not complaining anymore Jun 13 11:07:10 ok :) Jun 13 11:11:44 is procd_add_network_trigger still wip? Jun 13 11:23:11 aparcar: was it ever wip? Jun 13 11:24:07 do you mean "procd_add_interface_trigger" ? Jun 13 11:28:59 jow: I checked the (outdated) wiki and couldn't find it in the packages repo so figured it's not used https://wiki.openwrt.org/inbox/procd-init-scripts Jun 13 11:29:52 but procd_add_interface_trigger seems to be exactly what I was looking for, thank you very much Jun 13 11:41:32 so, have you updated the current wikiw ith your findings? Jun 13 11:43:51 karlp: I guess this is the to be extended version? https://openwrt.org/docs/guide-developer/procd Jun 13 11:43:56 If so, I'll update it Jun 13 11:57:03 I would have gone with https://openwrt.org/docs/guide-developer/procd-init-scripts fist, but either :) Jun 13 12:33:02 hmmmm Jun 13 12:33:56 https://pastebin.com/raw/kL3ifZMC Jun 13 12:35:40 https://pastebin.com/raw/1BDxK7L0 Jun 13 12:35:53 what you guys think? Jun 13 12:42:44 https://www.linuxquestions.org/questions/programming-9/undefined-reference-to-readline-494533/ Jun 13 12:43:02 the comments from that thread could apply Jun 13 12:45:09 may be a change in the package I gues Jun 13 12:46:02 either that or its a change in readline 7.0 Jun 13 13:03:41 ah cool, hadn't pulled from master for a while :D Jun 13 13:03:50 also just didn't build gdb heh Jun 13 13:19:20 recently updated my NBG6817 to a snapshot version. Worked very good, but now the "comp-lzo" option for openvpn is lost between /etc/config and the file generated for openvpn daemon.. Jun 13 13:19:41 is this a packaging issue? or luci? don't know where to report.. Jun 13 13:21:28 morning Jun 13 13:22:10 * ldir is selling packets of gravel & false beards Jun 13 13:30:21 * karlp laughs Jun 13 13:30:36 karlp: fancy a haggle? Jun 13 13:31:40 * ldir throws in some ppp packet dumps for free Jun 13 13:43:39 did anybody here run into the conntrack entry leak issue with the flow offloading code? Jun 13 13:43:47 if so, please update to the latest version and test Jun 13 13:43:50 it should be fixed now Jun 13 14:18:57 nbd: wow, nice work! Jun 13 14:19:31 what was the cause? Jun 13 14:19:45 something wrong with the tuple hashes? Jun 13 14:20:00 it was a race condition Jun 13 14:20:15 conntrack core bumping ct timeout in the gc worker for offloaded entries Jun 13 14:20:32 when a connection teardown is requested, it is left in the hash table and the offload bit is still set Jun 13 14:20:42 until the flow offload gc step cleans it up Jun 13 14:20:44 I should really test some offload Jun 13 14:21:00 so if the connection was bumped to slow path and not yet cleaned up, the ct core would bump the timeout to 24 hours Jun 13 14:21:13 i moved the timeout bump to the flow offload core Jun 13 14:21:18 and made it check for teardown first Jun 13 14:22:10 i'll wait for feedback from some people before i push the change to 18.06 Jun 13 14:37:15 linux command: traceroute -w 1 -T -A -p 80 IP - ok, openwrt the same - nothing. openwrt traceroute can not -T TCP? Not good. udp and icmp, not always interesting, udp and icmp not always the same trace as tcp Jun 13 14:38:08 traceroute in openwrt not as traceroute in linux? Jun 13 14:56:57 mmm Jun 13 14:57:29 so whats the status of bridge accel? we've got -j OFFLOAD, what can we do about software bridged stuff? (things like vxlan etc I guess) Jun 13 15:22:38 shouldn't the procd service file of cron use `procd_set_param file /etc/crontabs/root` to automatically reload on changes? Jun 13 15:57:12 qwefytuiityty: no, openwrt traceroute is busybox traceroute which has less features Jun 13 15:57:25 qwefytuiityty: I suggest "mtr" as replacement tracing utility Jun 13 15:58:08 aparcar: yes, actually it should register each file it finds in /etc/crontabs/ Jun 13 15:58:16 not just root Jun 13 15:58:25 you could e.g. have an /etc/crontabs/nobody too Jun 13 16:04:20 Hi, I'm trying to re-add the mysql package (instead of the current mariadb - it's too large) Jun 13 16:05:07 two issues: Jun 13 16:05:34 mariadb is still building as hostpkg (is it meant to?!) Jun 13 16:06:00 packages with mysql dependencies do not build as they claim a missing dependency: Jun 13 16:06:08 Package php7-mod-pdo-mysql is missing dependencies for the following libraries: libmysqlclient.so.18 Jun 13 16:06:12 any ideas? Jun 13 16:13:03 jow: should o create a PR? Jun 13 16:14:18 ping blogic, ping hauke Jun 13 16:14:56 playing with an Lantiq AR9 device (Fritz!Box 7330 SL) and wonder how etop could ever have worked with that board Jun 13 16:15:39 it got 1xAR8035 (LAN1) and 1xAR8030 (LAN2). the AR8035 is probed during boot and packets from that port arrive on eth0 Jun 13 16:15:48 sending anything on that port fails though Jun 13 16:16:44 ltq_etop_gbit_init() obviously doesn't initialize the second port and also lacks register definitions to easily add that Jun 13 16:19:04 i reckon there are bits for tx_delay in LTQ_GBIT_RGMII_CTL, with that set properly at least one of the two ports could work Jun 13 16:19:16 so: Lantiq AR9 datasheet, anyone? Jun 13 16:20:10 Haha, datasheets Jun 13 16:20:45 jow ned trace tcp, mtr trace tcp? And trace rout in linux this not the same as mtr. Jun 13 16:20:51 need Jun 13 16:21:11 traceroute Jun 13 16:22:01 qwefytuiityty: mtr defaults to icmp but can use udp (--udp) or tcp (--tcp) instead Jun 13 16:22:23 I know that mtr is not tracerouter, I was merely suggesting an opkg-installable alternative Jun 13 16:23:01 it is better, faster and more versatile than traceroute for all practical purposes (imho) Jun 13 16:23:38 me need mtr and traceroute as linux, not only any programs trace ip Jun 13 16:27:13 i heve linux in virtual machine, but if openwrt have traceroute as linux me not need for traceroute start virtual machine with linux. Jun 13 16:27:40 ssh Jun 13 16:28:04 putty Jun 13 16:40:46 only thinkg I don't like about mtr is that it clears the screen on exit :) Jun 13 16:50:07 Hi sqm is not working verry good for me on my wrt3200acm Jun 13 16:50:32 Should the Interface name be set to ath1.2 or ath1 Jun 13 16:55:44 neither Jun 13 16:55:50 guys I'm maybe a bit offtopic but I have a question to ask. How does the kernel load a firmware after a request to userspace? Who is in userspace that load the firmware for the kernel? Jun 13 17:22:39 mangix what should it be set to then? Jun 13 17:22:58 My wan on the luci stats page says ath1.2 Jun 13 17:23:49 It says Device: Software VLAN: "eth1.2" Jun 13 17:28:44 Hi Borromini Jun 13 17:29:12 Borromini you know on the wrt3200acm what should i have my interface set to under SQM? Jun 13 17:29:37 for my IP4 upstreem says Device: Software VLAN: "eth1.2" Jun 13 17:30:11 so should i have it set to ath1.2? Jun 13 17:30:12 Tapper: i don't think you should use software interfaces with sqm Jun 13 17:30:21 but i'm no authority Jun 13 17:31:08 I have had it set that way for a long time but just notest that my sun is downloading a game on the xbox and the net went to shit Jun 13 17:32:06 should i set it to eth1 then? Jun 13 17:32:36 hitech95: usually the kernel emits a firmware uevent which is handled by userspace. On linux that'd be udev (systemd), on openwrt procd (formerly hotplug iirc). Jun 13 17:32:44 You see I ran tests with dsl reports and all looked good Jun 13 17:33:27 hitech95: procd encodes a rule in /etc/hotplug.json which states: [ "if", [ "has", "FIRMWARE" ], [ ... [ "load-firmware", "/lib/firmware" ], ... ] Jun 13 17:33:43 would it help if I show you my /etc/config/network file and my /etc/sqm file? Jun 13 17:34:03 jow, so systemd or procd take the firmware name from the event FIRMWARE variable and pass look on /lib/firmware and send that via sysfs back to the kernel? Jun 13 17:34:13 hitech95: "load-firmware" is a builtin action, leading to https://git.openwrt.org/?p=project/procd.git;a=blob;f=plug/hotplug.c#l335 Jun 13 17:34:50 hitech95: handle_firmware() in turn indeed uses sysfs pseudo files to transfer the firmware image Jun 13 17:35:26 jow, thnx now I have finnaly understand how the phy firmwares on lantiq are sent to the kernel. :) Jun 13 17:36:03 *thanks *finally Jun 13 17:40:47 jow do i have my sqm set rite? Jun 13 18:11:17 hello Jun 13 18:11:48 i was following this wiki guide Jun 13 18:11:49 https://openwrt.org/docs/guide-user/luci/getting-rid-of-luci-https-certificate-warnings Jun 13 18:12:11 to get openssl certs Jun 13 18:12:29 but i am getting an error running openssl req Jun 13 18:12:35 can anybody help me ? Jun 13 18:19:08 dwmw2_gone: apparently the problem with openzwave control panel is that the html/js files are gzipped Jun 13 18:19:21 dwmw2_gone: if I gunzip cp.html.gz the control panel starts working again Jun 13 18:27:30 o/ stintel Jun 13 18:27:56 qtfy: do post your error instead of waiting for someone to offer you help with a problem you are not detailing. Jun 13 18:59:20 Borromini nevermind i got it ! Jun 13 19:03:35 I am trying to "port" openwrt for the Tp-Link Archer D7 using an initramfs image before overwriting the OEM firmware, but I am stuck with this error: "Not enough buffer for decompression LZMA ERROR 1". Has anyone run into the same problem? Jun 13 19:04:55 I have also created this thread to recap my tests https://forum.lede-project.org/t/uboot-not-enough-buffer-for-decompression-lzma-error-1/ Jun 13 19:49:18 can anyone explain please what this c means: pvcc->encaps = (enum pppoatm_encaps) be.encaps; Jun 13 19:49:37 particularly the (enum blah) bit Jun 13 20:15:43 Tapper, you wrote ath Jun 13 20:16:05 Yeah sorry I ment eth Jun 13 20:16:51 it shouod be the interface that maps to your WAN connection Jun 13 20:17:25 I get that but on the luci page it says Software VLAN: "eth1.2" Jun 13 20:17:39 sounds right Jun 13 20:17:51 I read that sqm is not to be used on vlans Jun 13 20:18:07 ah slightly incorrect Jun 13 20:18:30 Why is my wan listed as a Software VLAN Jun 13 20:18:31 erm. actually i'm confusing myself Jun 13 20:18:43 because it is Jun 13 20:19:18 your router has all 5 ports switched Jun 13 20:19:32 is it ok to still use sqm on eth1.2 then Jun 13 20:19:33 it's not 4 + 1 Jun 13 20:19:40 shouod be Jun 13 20:19:55 what's your WAN speed? Jun 13 20:20:08 200 down and 14 up Jun 13 20:20:34 With no sqm i get a F for bufferblat Jun 13 20:21:22 one option to keep your sanity is to switch to DSA. Speed goes from gigabit to 500mbps from WAN to LAN but should be fine for you. Jun 13 20:21:37 dsa? Jun 13 20:23:05 different switch driver Jun 13 20:23:47 VLANs go away. You only get a wan interface and lan[1-] Jun 13 20:23:52 1-4 Jun 13 20:24:24 I don't use vlans anyway Jun 13 20:24:42 How do you use dsa? Will have to build? Jun 13 20:24:53 needs to be patched in Jun 13 20:27:49 I mite just flash back to a newtown build. I am running a plane snapshot build now Jun 13 20:29:51 I am going to see what bufferblat is like when using flowoffload Jun 13 20:33:24 score should improve Jun 13 20:33:54 Does flowoffload not interfere with QoS? Jun 13 20:36:52 qos is off and flowoffload is on now and bufferblat is a B so I get A+ withsqm but 178 down and a B with flowoffload and 218.5 wtf Jun 13 20:38:58 My last sentence did not make that mutch sense sorry. Jun 13 20:39:38 with SQM I get A+ for bufferbloat 175 down and 10 up Jun 13 20:40:21 with flowoffload I get a B for bufferbloat with 218.5 down and 12.8 up Jun 13 20:41:21 with no sqm or flowoffload I get a F for bufferblat with 218 down and 12.7 up Jun 13 20:42:27 Interesting that flowoffload improves the bufferbloat. Jun 13 20:42:45 yeah that'swhat I was thinking! Jun 13 20:43:25 * Tapper thinks that my space bar is on it's way out! Jun 13 20:44:44 mamarley: hardware not software Jun 13 20:44:56 hardware bypasses the qdisc layer Jun 13 20:45:14 Ah, OK, that makes sense. Jun 13 20:46:40 Tapper: makess some sense. SQM is more aggressive about throttling speeds to reduce bufferbloat Jun 13 20:49:24 there was an explanation for the lower bufferbloat with flowoffload but I don't remember where I saw it Jun 13 20:59:55 so, if I'm trying to work out where the rt28xx caldata partition is, does anyone have an example of what it should look like? Jun 13 21:00:02 is there a signature or anythign I can try looking for? Jun 13 21:00:42 I've got a flash dump, but the broadcom imagetag partition parser only gave me CFE,nvram,linux,rootfs,kernel. Jun 13 21:15:05 the bcm partition sorry is just three, CFE for 128K, "linux" for 16128K, then nvram for another 128K, making up the full 16M. Jun 13 21:24:05 so.. looks promising, even if binwalk didn't notice it. Jun 13 21:31:06 karlp: have you found it? if not, I could provide boardconfig for the VGV7510KW22/ lantiq (RT3062F) Jun 13 21:31:13 https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/lantiq/files-4.14/arch/mips/boot/dts/VGV7510KW22.dtsi;h=ab558372ee03a11a206c7cbcac42cbb32f4f94a9;hb=HEAD#l236 Jun 13 21:32:51 worth a shot? Jun 13 21:32:58 I'm looking for what corresponds to https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/brcm63xx/patches-4.9/422-BCM63XX-add-a-fixup-for-rt2x00-devices.patch;h=d021b369046a6c37686bc61b7f8327f9dd86d142;hb=HEAD#l54 Jun 13 21:33:16 I've got no caldata defined at the moment, and it detects the device, lspci shows it, it just doesn't work, Jun 13 21:33:23 http://paste.debian.net/1029128/ Jun 13 21:33:47 but the other devices in bcm63xx all have the cal_data in the last 128K of the linux partition, just before NVRAM, and my flash is all 0xff there. Jun 13 21:34:06 12345678? nice signature :) Jun 13 21:35:23 interesting that it starts with 0x30 0x62 at 0x410, for a RT3062F Jun 13 21:38:32 do you find strings like bb_m1_thres, bb_m1_thresh_low, bb_m2_thresh_low somewhere? Jun 13 21:39:05 nope, nowhere. Jun 13 21:41:31 well, this paste.debian.net/hidden/7916a151/ is the complete dump for boardconfig Jun 13 21:42:20 yeah, ok, that's nothign like mine. Jun 13 21:42:25 perhaps it's not meant to have any Jun 13 21:42:41 quite possibly for PCI attached devices Jun 13 21:43:05 except that the brmc63xx board file is explicitly doing this caldata load based on the pci id :| Jun 13 21:43:19 it's optional, but still, Jun 13 21:50:15 http://paste.debian.net/1029134 interesting Jun 13 21:54:13 found someone else on the internet saying they suspect this lzw section is missing the dictionary. Jun 13 21:55:42 migth be barking up the wrong tree. Jun 13 21:55:55 enough for tonight. will try a different wifi debug approach another night Jun 13 22:10:26 so, you guys were talking about 10gig stuff the other day, is it still possible to buy new 100meg fibre NICs? or are they now unobtainum? Jun 13 22:11:26 I've got this device with a 100m fibre port on it, and I'd like to talk to it, as it's the original "wan" port Jun 13 22:20:04 sadly no fibre here (o.k., that's a lie, I have a 1000BASE-SX module in an ancient salvaged Alcatel OmniStack 6124 switch, but no opportunity to test it) Jun 13 22:21:33 I've found the _downstream_ one, that I already have, but I have to find the matching upstream one now :) Jun 13 22:47:35 build #886 of ar71xx/mikrotik is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ar71xx%2Fmikrotik/builds/886 Jun 13 23:36:46 ping neheb Jun 14 00:07:18 hmm 17h and 148mb of dropped arp packets. that would be 6268mb in 30 days? even for docsis that seems quite high for me Jun 14 00:29:22 I heard somebody works on something called mdns to have a smaller dnsmasq alternative, is that so? On git.openwrt I can't find any recent changes, could somebody please point me in a direction? Jun 14 00:44:05 build #925 of x86/legacy is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/x86%2Flegacy/builds/925 Jun 14 00:58:17 aparcar: mdns is not an alternative to dnsmasq. Jun 14 01:24:36 drmr: okay please excuse the confusion Jun 14 01:25:46 aparcar: no worries. it's a legitimate question. mdns just doesn't do the same as dns. Jun 14 01:34:02 biangbiangmian: pong Jun 14 01:34:32 drmr so there are no such ambitions? Jun 14 01:34:45 aparcar: ambitions for what? Jun 14 01:35:45 a smaller dnsmasq alternative. I heard dangowrt talk about it Jun 14 01:39:40 aparcar: I believe to make a smaller "dnsmasq" you would have to strip out functionality. at which point it stops being an "alternative." Jun 14 01:39:55 aparcar: at which point you would have to discuss what to keep and what to strip. Jun 14 01:40:11 @mangix ah the legend himself Jun 14 01:40:55 :) Jun 14 01:41:22 lol Jun 14 01:41:25 aparcar: also, dnsmasq that ships with openwrt is already stripped down. Jun 14 01:42:22 drmr ack, thanks :) Jun 14 01:42:46 been playing around with this .dts stuff, but it's probably fairly obvious i don't really know what i'm doing lol Jun 14 01:44:04 that reg entry, what's that? is that the address where the wifi chip's eep rom starts? Jun 14 01:46:22 https://github.com/devicetree-org/devicetree-specification/blob/master/source/devicetree-basics.rst#reg Jun 14 01:47:10 that reg setting should not be touched I think Jun 14 01:49:09 ah thanks for that, been looking for something like this Jun 14 01:50:19 so it is the offset of where the eeprom for the wifi chip is, within the SoC's memory Jun 14 01:50:48 aparcar: there are always 'ideas' to replace dnsmasq, but it usually doesn't go far beyond the idea phase - because it's hard to find a replacement that really covers its use cases and isn't much larger. the dhcp parts are 'easy', the DNS functionality is another story (and highly security sensitive), not even to start about dnsmasq's additional features like bootp et al Jun 14 01:52:17 biangbiangmian: a subsection of that. keep in mind the mt76 node is within a pcie node Jun 14 01:52:35 anyway, i wouldn't touch dts files Jun 14 01:53:06 it seems to be the only thing that gets the wifi working, what else could i do? Jun 14 01:53:39 when did wifi break? Jun 14 01:55:19 i don't have it pin pointed to when it originally broke because i was on a snapshot for a while Jun 14 01:55:55 i just remember that an earlier 4.9 build was working well Jun 14 01:56:06 around november last year Jun 14 01:56:41 https://forum.lede-project.org/t/unielec-u7621-mediatek-mt7621-mt7612-mt7602-wifi/8833/6 Jun 14 01:56:41 are you building master or 18.06? Jun 14 01:56:52 i've tried both master and 18.06 today Jun 14 01:58:11 aparcar: the usual complaints about dnsmasq are about it being 'hard' to integrate (that means it needs lots of restarting, instead of more dynamic config changes underneath) and it being not that easy to get changes upstream, not really about its reliability or functionality itself - and it is battle tested widely, getting any other potential alternative up to its stability level takes real efforts Jun 14 01:58:17 and time Jun 14 01:59:12 biangbiangmian: i see potential problems with master but not with 18.06 Jun 14 02:06:27 what kind of problems in master? Jun 14 02:08:47 my dts cleanup patch is untested Jun 14 02:27:19 pkgadd thanks for the clarification. I heard that some OpenWrt devs work on a new DHCP/DNS system but that may just been a missunderstanding Jun 14 02:36:31 well, when i use nothing in the .dts file except for "status = "okay"", i can enable 2.4, but not 5ghz **** ENDING LOGGING AT Thu Jun 14 03:00:04 2018