**** BEGIN LOGGING AT Wed Jun 27 03:00:01 2018 Jun 27 03:21:23 WPA3 was announced yesterday. What does this mean for OpenWRT? https://www.wi-fi.org/news-events/newsroom/wi-fi-alliance-introduces-wi-fi-certified-wpa3-security Jun 27 03:22:14 nothing Jun 27 03:22:41 the announcement doesn't really contain any meat Jun 27 03:23:05 so it's still not really known what to expect from a compliant WPA3 client or AP Jun 27 03:23:59 nor is it clear if a version of upstream linux (mac80211/ wlan drivers/ firmware) or hostapd exist so far that do support it Jun 27 03:24:29 it just needs time for this stuff to materialize in practice, not any paper launch Jun 27 03:24:50 Makes sense. Thanks. Jun 27 03:28:13 even the big name vendors don't expect to ship anything WPA3 before next year (and I wouldn't expect the first half of 2019 either) Jun 27 03:28:55 those who are members of the WiFi-Alliance™ and part of all the relevant standardization groups guilty of this new standard Jun 27 03:29:10 the main problem with current WPA2 being unpatched clients from KRACK ? Jun 27 03:29:31 hard to see what WPA3 really fixes in that case Jun 27 03:30:32 one of the things they said WPA3 would fix, is "open" wifi but that is still encrypted properly. Jun 27 03:30:33 it makes a few existing things (semi-)mandatory, while at the same time not really offering anything new when being run in WPA coexistance mode (which will remain a necessity for at least 5 years to come) Jun 27 03:31:23 someone smarter than me commented on WPA3: "wpa_supplicant supports all of it afaik" Jun 27 03:31:56 heck, I still can't make ieee 802.11w mandatory in my network (while I have enforced WPA2 from early on) Jun 27 03:32:42 yeah that's always a hassle Jun 27 03:32:53 even if it works it seems to rape performance most of the time Jun 27 03:33:27 performance isn't an issue at all, just not all devices having support for it (at all) Jun 27 03:33:54 but yes, I've heard that not all drivers work well with 802.11w enabled Jun 27 03:54:33 wilson5420 WPA3 will use different encryption for each client. Jun 27 04:11:40 is is possible to operate on a list with logic operations, say I have a list of IP I want to ! on them together(without ipset, as IP is just an example) Jun 27 04:12:27 something like 'iptables -A INPUT ! (src ip1 | src ip2 | src ip3) REJECT' Jun 27 04:52:31 ausjke: no, this is only possible with either ipsets or extra chains Jun 27 04:53:48 ausjke: something like iptables -N filter_list; iptables -A INPUT -j filter_list; iptables -A filter_list -s ip1 -j RETURN; ipables -A filter_list -s ip2 -j RETURN; iptables -A filter_list -j REJECT Jun 27 05:06:33 FS#1612 seems interesting Jun 27 05:07:52 Tapper: its one of these tickets where I as a developer can simply shrug Jun 27 05:08:08 thei ssue is there on the vendor firmware as well, what's he expecting us to do? Jun 27 05:08:16 rewrite the mwlwifi driver? Jun 27 05:09:01 he says that he is running 18.06 Jun 27 05:09:22 he also writes that the vendor firmware has the same jitter problem Jun 27 05:09:57 But wouldent trafic from out side the network be inspected by the firewall macking it slower Jun 27 05:14:57 I don't think he did any testing over wi-fi as at thend of the bug he rites "wifi stations due to inactivity after very short periods of time, but that is outside the scope of this bug " Jun 27 05:15:27 I think all his testing was over Ethernet Jun 27 05:15:29 *shrug* Jun 27 05:15:34 lol Jun 27 05:16:01 I am on same ISP but would not know were to start testing that out Jun 27 05:16:15 I have wrt3200acm to Jun 27 05:16:27 maybe it is sqm related Jun 27 05:16:32 and same speed connection Jun 27 05:16:44 same hub to in same moad Jun 27 05:18:47 If i was to give you my IP would there be a way to test? Jun 27 05:19:02 I can't help with that Jun 27 05:19:12 k Jun 27 05:19:18 quit Jun 27 05:19:27 maybe it is some problem with the ethernet driver Jun 27 05:19:35 but then I wouldn't know what to fix there Jun 27 05:21:36 Do you know any one who would? Jun 27 05:25:12 Hi rmilecki Jun 27 05:25:28 Tapper: hi Jun 27 05:26:10 I was just asking about FS#1612 do you think there is a good way to test? Jun 27 05:26:35 I am on same isp with same set up but it is the wrt3200acm not the x Jun 27 05:27:37 I don't have a way to test from out side my network Jun 27 05:27:38 Tapper: why asking me in particular? Jun 27 05:27:56 Just picked you because you joiend Jun 27 05:28:04 ... Jun 27 05:28:06 no comments Jun 27 05:28:24 .. and friendly? :) Jun 27 05:28:28 lol Jun 27 06:49:33 hmm, why on earth keep people inveinting pacakging systems which are unable to clean up after themselves Jun 27 06:49:45 * Tapper thinks how can it be just 7.49 and he is board! Jun 27 06:50:03 first npm, which killed our builders by filling /tmp with millions of temporary directories Jun 27 06:50:18 then jsmake, which did the same with millions of minified .js files Jun 27 06:50:40 now I discovered that LuaRocks does the same, dropping millions of temp directories in /tmp which are never purged Jun 27 06:50:42 nmp, jsmake... JavaScript world Jun 27 06:50:45 what did you expect Jun 27 06:51:20 i had no idea we use nmp on builders Jun 27 06:51:26 we use nodejs Jun 27 06:51:31 uh Jun 27 06:51:34 or rather we build packages that use nodejs Jun 27 06:51:57 thats the problem with processing package feeds, you never know what kind of crap you have to process Jun 27 06:52:20 I should probably start working on throwaway containers to reset the system after each build Jun 27 06:52:34 fine new IT world Jun 27 06:53:06 because stuff craps all over the place, the new normal is to use "containers" which you can simply dispose because noone can guarantee any kind of consistent system state anymore Jun 27 06:54:15 yeah, that sucks Jun 27 06:55:18 i sincerely hate npm & all that JavaScript compilation Jun 27 06:55:33 i tried once to "build "some silly tiny projects for popups from the source Jun 27 06:55:48 i gave up due to amount of dependencies, time & space needed Jun 27 07:00:27 buildbot@builds:~$ find /tmp -maxdepth 1 -iname 'luarocks*' | wc -l Jun 27 07:00:29 5604 Jun 27 07:01:18 need to review that package eventually to check if it uses $TMPDIR or harcoded /tmp Jun 27 07:01:38 because the former is supposed to point into the buildroot tmp/ which is frequently cleared Jun 27 07:21:30 I swear there is this fly in my house and it keeps buzzing me in the forehead! Jun 27 07:21:46 It's seding me around the bend! Jun 27 07:41:52 My 11 year old sun killed the fly. Good kid! Jun 27 07:42:48 Tapper: reminds me of the karate kid movie Jun 27 07:42:55 did he use chop sticks by any chance ? Jun 27 07:43:03 bright lad ;-) Jun 27 07:43:05 that we flipping awesome Jun 27 07:43:21 haha yeah at one point he did want to slap it on my head! Jun 27 07:43:39 not sure if that would be good or bad Jun 27 07:43:43 I told him no but if he gets it i will get him a bag of sweets after school Jun 27 07:44:01 Can anyone backport this change to openwrt-18.06 branch, please ? https://github.com/openwrt/packages/pull/6062#issuecomment-400481991 Jun 27 07:51:08 sooo all network interfaces stopped responding to traffic again Jun 27 07:51:45 ifup lan fixed it again Jun 27 07:52:16 maybe next time I should try fw3 reload/restart instead, see if that helps Jun 27 07:52:43 but this is an office with ~60 people working so not really a lot of time to dig deep Jun 27 07:52:45 stintel: anything in logread/dmsg wrt. carrier Jun 27 07:52:49 ? Jun 27 07:52:50 jow: nothing at all :( Jun 27 07:53:10 ok so no l3 / netifd ip config issue Jun 27 07:53:54 anything indicated by the blickenlights (if you have access?) bridge/interface gone into some mad loop? Jun 27 07:54:17 ldir: well all interfaces I tried stopped responding Jun 27 07:54:35 this is a watchguard xtm505 with 6 e1000e interfaces Jun 27 07:54:51 no response to traffic on wan1/wan2 nor lan Jun 27 07:55:04 and only lan is bridged Jun 27 07:55:39 yesterday I unplugged and replugged the LAN UTP, saw that in dmesg, no change Jun 27 07:56:17 reminds me on some of blogic's war stories with e1000 in the laptop Jun 27 07:57:51 and actually I think I have seen such behavior before Jun 27 07:58:14 here it's not the 3rd time in 24h Jun 27 07:58:16 now* Jun 27 07:58:31 hmm, recently I had an issue with my main archer & slave AP going missing and the symptoms basically were an uplink interface going 'mental' - replug that and all came back to normal. Jun 27 07:59:35 I forced unique macs onto each bridge over the vlan uplink port and so far (1 month) not seen a repeat. Jun 27 08:06:37 and yes I tried STP Jun 27 08:06:47 stintel: rps/xps ? Jun 27 08:07:09 stintel: let me dig in my memories Jun 27 08:07:25 basically can make e1000 deadlock with too much traffic Jun 27 08:07:50 while doing the offloading testing i am passing gigabytes of 128 byte frames Jun 27 08:08:03 and i can kill e1000 with 500k fps easily Jun 27 08:08:21 using rps/xps fixes the issue Jun 27 08:08:25 affinity made it worse Jun 27 08:09:00 and phillip prinvedille also did a lot of debugging of such issues Jun 27 08:20:06 netifd: remove rps/xps configuration support Jun 27 08:20:07 hmmm Jun 27 08:23:15 blogic: would enabling RPS/XPS make a difference on a single-core/single-threaded CPU ? Jun 27 08:23:28 no Jun 27 08:23:50 there is also a packaet aggregate feature for e1000e Jun 27 08:24:00 i think that bit pprinvidelle Jun 27 08:24:16 philipp64: ping Jun 27 08:24:33 philipp64: what was your story regarding e1000 again ? Jun 27 08:24:38 anyway I am just going to flash a new image later tonight Jun 27 08:24:40 i thin dgolle even helped him with it Jun 27 08:25:05 it's 4.14.37 / master r6884 Jun 27 08:25:59 going to make an image from 18.06 branch and flash that Jun 27 09:26:49 any idea why wan1 (eth6) and wan2 (eth5) are using different interrupts while lan (eth1) doesn't? https://paste.pound-python.org/show/AVTRKpCLerkIrZ6HFX7o/ Jun 27 09:29:48 stintel: eth1 is (renamed) onboard LAN (ICH7)? Jun 27 09:29:54 oh no Jun 27 09:29:58 * f00b4r0 spots eth0 Jun 27 09:30:15 dmesg might have an explanation Jun 27 09:30:19 yeah eth0 isn't used because slow Jun 27 09:30:32 I'd have to reboot the thing which is not an option right now :( Jun 27 09:32:48 most of my machines have tg3 hw so can't help there sowwy :P Jun 27 09:33:37 np Jun 27 09:38:31 eth1 is bridged, others aren't, maybe that's why Jun 27 09:41:33 quite possible indeed Jun 27 09:49:20 f00b4r0: hmm, on an identical device with different config (eth0/1/2/3/4 bridged) eth2/3/4 use 3 interrupts but eth1 again 1 Jun 27 09:49:22 weird Jun 27 09:50:52 indeed. I'm not familiar with that hw so can't help much i'm afraid Jun 27 09:52:40 stintel: Presumably the split queues are configurable - perhaps BIOS bug Jun 27 09:54:50 Can anyone backport this change to openwrt-18.06 branch, please ? https://github.com/openwrt/packages/pull/6062#issuecomment-400481991 Jun 27 09:56:53 I want to stop openwrt from running an interface reset which is triggered by /lib/preinit/preinit_ip_deconfig. Simplify removing the file stops openwrt from executing " ip -4 address flush dev $pi_ifname" but this may is an ugly solution. I'm trying to setup a Docker image for CI tests running OpenWrt. As most services require ubusd and procd I'm doing this wired stuff Jun 27 09:57:44 Can I simply tell openwrt to ignore some interfaces completly and keep them as they are? Jun 27 10:05:04 stintel: Can you paste a fully verbose lspci from that host? Jun 27 10:06:03 Monkeh: sure Jun 27 10:06:48 Monkeh: https://gist.github.com/stintel/efb8dd25b2908b4798e41ce230053e21 Jun 27 10:08:08 stintel: Thought so - https://gist.github.com/stintel/efb8dd25b2908b4798e41ce230053e21#file-stdin-L463 Jun 27 10:08:16 No MSI-X, no multiple interrupts Jun 27 10:09:06 I see Jun 27 10:09:32 Why that's disabled is between you and the vendor Jun 27 10:09:49 yeah I will just switch it to eth2 ;) Jun 27 10:10:10 this is a watchguard device, don't expect any support from the vendor Jun 27 10:10:27 heh Jun 27 10:10:33 You might be able to force it Jun 27 10:10:36 I've done worse Jun 27 10:10:43 but would it matter much to use different irq's for rx/tx queue? Jun 27 10:11:00 it's a single-core/single-threaded CPU anyway Jun 27 10:11:37 I doubt it would on that Jun 27 10:12:30 stintel: One of my boards has this in grub.cfg: write_dword 0xFED1F100 0x42210 Jun 27 10:12:30 setpci -d 8086:2828 90.b=40 Jun 27 10:12:34 Working around vendors is always fun. Jun 27 10:14:10 stintel: Single core single thread with a 4-series - Atom? Jun 27 10:15:13 Although they're mostly HT as I recall.. Jun 27 10:17:10 Monkeh: no Jun 27 10:17:16 model name : Intel(R) Celeron(R) CPU 440 @ 2.00GHz Jun 27 10:17:31 ew. Jun 27 10:17:37 didn't even know there were 64bit single-core CPUs Jun 27 10:17:44 well, x86 that is Jun 27 10:18:53 Er, the first ones were single core Jun 27 10:19:11 Back when AMD invented it and Intel were left looking silly. Jun 27 10:20:11 Pentium 4 Nocona Jun 27 10:20:22 Do not speak that name. Jun 27 10:20:38 it which shall not be named Jun 27 10:20:42 it wasn't that bad. served me nicely during the winter when i had to update gentoo :v Jun 27 10:20:55 no heater needed, right Jun 27 10:21:38 and actually comparing, i think haswell is hotter Jun 27 10:21:52 (not counting the perf/watt thingie) Jun 27 10:22:19 Anyway, nocona had HT Jun 27 10:22:20 since someone at intel had the brilliant idea of putting VRMs on the chip Jun 27 10:22:44 There were a few Prescotts with 64 and no HT, the evil useless things Jun 27 10:23:20 it wasn't that long ago when I thought bulldozer wasn't bad.. then rust came out and now a firefox update takes hours Jun 27 10:23:32 Monkeh: no iGPU, no ME, not THAT evil Jun 27 10:23:57 movi: No performance, no utility Jun 27 10:23:58 :P Jun 27 10:24:12 Monkeh: what would you call the RISC-V people then? :P Jun 27 10:24:30 Optimistic? Jun 27 10:25:18 They're pretty neat for soft cores, anyway Jun 27 10:25:30 soft cores? Jun 27 10:25:50 In FPGAs. Jun 27 10:26:03 ah Jun 27 10:26:09 There's no useful real hardware yet - small processors are all about the peripherals and they don't have any Jun 27 10:27:04 i thought you meant as a virtual ISA, in which case we already did Java ;) Jun 27 10:27:10 movi: RISC-V is apparently already in fairly extensive use as soft cores at CERN Jun 27 10:27:22 They appreciate being able to fully validate it. Jun 27 10:27:25 Monkeh: can't say, haven't worked there for 10 years Jun 27 10:28:06 Monkeh: during my tenure however, they were pissed that they were basically stuck with Pascal, since no newer library gave them "the correct results" mathematically Jun 27 10:28:58 i'll look into it Jun 27 10:29:01 Perhaps they should check their math :D Jun 27 10:29:42 >T. Włostowski Jun 27 10:29:48 of course it's a Polak :D Jun 27 10:30:46 Probably an RPN fan, too. ;) Jun 27 10:31:40 I'm Polish and I can't understand RPN. Guess I'm not Polish enough :< Jun 27 10:32:08 hmm, the paper is actually pretty rad Jun 27 10:40:10 yo, with the upcoming WPA3 standard, will the existing hardware support this mode (in terms of a firmware upgrade) or will a hardware upgrade be required? Jun 27 10:42:46 aha, starting to get setpci Jun 27 10:43:00 airwind: nobody knows Jun 27 10:43:46 there was some media hype a few months ago when it was announced that wpa3 will happen Jun 27 10:44:07 apparently there's been another media hype a few days back based on a press release of the wifi alliance stating that wpa3 will happen Jun 27 10:44:18 actual available information in both cases is zero Jun 27 10:44:34 but I guess I need setpci before loading the driver Jun 27 10:46:43 it is quite likely that modern softmac radios will implement most (all) of wpa3 in their firmware blobs Jun 27 10:47:07 ok, guess we'll just have to wait and see Jun 27 10:47:19 it is also quite likely that this will be a good opportunity to get rid of these pesky oss drivers Jun 27 10:49:29 here we go again... Jun 27 10:57:41 heaaaaapppps of commits in wpa_supplicant for wpa3 stuff fwiw Jun 27 10:58:34 airwind: I have to revise my comment, there is information... apparently one can download the spec document after registration Jun 27 10:58:39 I didn't bother to do that though Jun 27 10:59:35 Not going to be relevant to me until devices support it.. Jun 27 10:59:47 I can't even get .11w support out of a brand new phone Jun 27 11:00:30 Monkeh: but does that brand new phone have a recent OS version Jun 27 11:00:39 Adequately. Jun 27 11:00:50 It's newer than my ancient Kindle. Which supports .11w. Jun 27 11:00:55 but yeah 11w support is terrible, networkmangler only implemented it a year ago or so Jun 27 11:01:14 stintel: They claim to support it but so far I haven't been able to turn it on via any UI Jun 27 11:01:16 hi all, my asus rt-n56u on o/l 17.01.4 keeps dropping wifi. It's a standard issue since forever, v14, where rt2xx kernel modules fail with all sorts of errors. Only reboot is possible after. So i'm wondering, is there any news about this? Jun 27 11:04:08 so how to run setpci before loading modules, in openwrt Jun 27 11:04:22 That I can't tell you Jun 27 11:04:24 stintel: preinit Jun 27 11:04:34 You figured out the magic to enable MSI-X and hope for the best? Jun 27 11:04:38 or coldplug Jun 27 11:04:57 Personally i tend to twiddle bits down in the bootloader Jun 27 11:06:49 ok so just a custom file in /lib/preinit Jun 27 11:07:26 prolly Jun 27 11:07:31 tha Jun 27 11:07:33 but beware, that is a compile thing Jun 27 11:07:43 coz preninit runs from r/o fs Jun 27 11:07:54 yeah, going to add it in my next image Jun 27 11:07:57 so adding a file to preinit squashfs wont work Jun 27 11:08:57 setpci -s 02:00.0 -d 8086:10d3 CAP_MSIX.L=80040011 Jun 27 11:09:09 funky Jun 27 11:11:01 blah, libwebsockets kneejerked and closed all their github because "we hatez microsoft" Jun 27 11:11:07 so now there's no issue tracker anymore. Jun 27 11:11:11 top job guys. Jun 27 11:11:24 "but we selfhosted ze git repo!" Jun 27 11:11:34 probably not going to fix my network problem but now at least all interfaces are going to be configured the same Jun 27 11:17:38 jow do you have an idea to setup preinit.conf ignoring a preconfigured interface? in this case eth0 Jun 27 11:23:41 stintel: May also set your interfaces on fire Jun 27 11:24:26 Monkeh: then I'll do a little firedance Jun 27 11:25:23 aparcar: no, I've never done that Jun 27 11:39:45 jow and can I manually start ubusd without running procd? This all may sound very odd, sorry Jun 27 11:50:51 just.... start it? Jun 27 11:50:53 what do you mean? Jun 27 11:51:09 you can run it on your laptop if yo ulike. Jun 27 11:53:04 karlp I'm trying to start it via `ubusd` but services still won't find it Jun 27 11:53:47 you probably have the wrong socket path for either the services or yoru invocation then? Jun 27 11:58:52 build #38 of rb532/generic is complete: Failure [failed dirupload] Build details are at http://release-builds.openwrt.org/18.06/images/builders/rb532%2Fgeneric/builds/38 blamelist: Alberto Bursi , INAGAKI Hiroshi Jun 27 14:31:13 if I have a remote configured in a git repo config thus "url = git://git.openwrt.org/openwrt/staging/nbd.git" what protocol is used to get the repo ? Jun 27 14:33:59 or the questions behind the question... git@server.com:userdir/repo.git is ssh? Jun 27 14:36:48 According to the manual, "git clone [user@]server:project.git" is ssh protocol Jun 27 14:36:48 https://git-scm.com/book/en/v2/Git-on-the-Server-The-Protocols Jun 27 14:37:25 luaraneda: that's without git:// at the front.... Jun 27 14:37:47 or, yeah, ldir's second one Jun 27 14:37:55 "git://git.openwrt.org/openwrt/staging/nbd.git" is using the git protocol, which is unauthenticated and read only (you can't push) Jun 27 14:39:40 If I remember correctly, the git protocol uses UDP, and is blocked by most corporate firewalls Jun 27 14:42:40 I had to send an e-mail to my university's firewall administrator requesting to open the port so I could clone a repository, and I had to explain them what is a repository and why I needed it... Jun 27 14:46:06 thanks guys - I think I'm hitting a server issue but it's really causing me to doubt myself :-) Jun 27 14:47:49 luaraneda: this is why any repohost worth it's salt supports http(s) git clones Jun 27 15:22:39 jow: thanks for the iptables help Jun 27 15:23:23 does ca-bundles include all the intermediate certificates? seems not, but could not find a place to download any "intermediate CA bundle" Jun 27 15:26:01 perhaps try stating what your actual problem is? Jun 27 15:31:40 just some forum posting mentioning it and was curious about it as I thought ca-bundle contains everything already Jun 27 15:42:42 until ca-bundle is renamed as "a few certs, not all, definitely not anything useful" you should perhaps question the reporter a bit more first... Jun 27 15:44:22 would expect chrome/firefox has a near-to-complete ca-bundle, I just use the ca-bundle package in openwrt which is from debian Jun 27 16:02:24 * ldir is living in a strange parallel universe where Germany are out of the world cup in the first round. What the hell is going on? Jun 27 16:02:29 is slack still dead? Jun 27 16:02:56 is what what? Jun 27 16:03:05 ldir: no more last 15 seconds miracle? Jun 27 16:03:38 ldir's british... Jun 27 16:03:42 and it's over now anyway Jun 27 16:03:45 ausjke: a last 15 second debacle rather :) Jun 27 16:03:52 congrats to sweden though Jun 27 16:05:09 ldir: that means that people fineally start working again here Jun 27 16:05:19 lol Jun 27 16:05:34 lazy twats blocking roads cheering at tv screens Jun 27 16:05:54 wait, you can always expect something from France when time comes, i mean they can both play and act well in the field Jun 27 16:06:05 * karlp laughs Jun 27 16:06:16 * ldir gets a slight hint that jow isn't a football fan :-) Jun 27 16:09:19 I'm a football fan Jun 27 16:09:34 When it's on everyone buggers off indoors and leaves the roads clear. Jun 27 16:12:04 * jow wonders how the public tv is going to fill up all these slots now Jun 27 16:13:39 just shows the same games to less people. Jun 27 16:28:10 Всем, привет! Есть кто живой? Jun 27 16:44:37 Wed Jun 27 18:43:39 2018 daemon.notice procd: /etc/rc.d/S19dnsmasq: Failed to parse json data: unexpected end of data Jun 27 16:44:41 any idea what this could be ? Jun 27 16:47:11 stintel: I'd say some of the procd routines Jun 27 16:52:54 I am installed lede 17.01.4 image in mikrotik metarouter. Ping command work. Try command "opkg update" and get error: Collected errors: * opkg_download: Failed to download http:/ es/Packages.gz, wget returned 8. --- Any idea how to fix it?! Jun 27 16:55:52 Monkeh: even with MSI-X enabled via setpci eth1 still uses only 1 IRQ Jun 27 16:56:26 Hm. Jun 27 16:56:31 There may be more to the setup Jun 27 16:56:38 I only skimmed the datasheet. Jun 27 17:04:28 stintel: pong Jun 27 17:05:01 blogic: was the e1000e, as I remember... the i82574 and it would lock up under heavy load. Jun 27 17:05:53 blogic: what bit me was that RPS is turned off by default. so I turned it on and my 10G throughput was much better. Jun 27 17:07:15 see https://github.com/openwrt/openwrt/commit/3a69ad3b2a0df578ca340ba95f35230ee9529c9c Jun 27 17:08:22 which was a redux of a0c05a7c94137945b93dacc2dcd0f27d09539650 Jun 27 17:09:18 stintel: what about MSI? Jun 27 17:10:39 does this box ever support MFD's (multi-function devices)? If so, you probably need MSI. If not, you can use MSI-X. Less overhead but also less flexibility. Jun 27 17:11:55 philipp64: did you see https://gist.github.com/stintel/efb8dd25b2908b4798e41ce230053e21 Jun 27 17:12:46 looking now. Jun 27 17:15:31 okay... looks like it does better with MSI-X than MSI, so it needs to be reconfigured in BIOS. Jun 27 17:16:59 02:00.0 - 07:00.0 I mean... not sure what 08:00.0 is... Jun 27 17:18:13 looks like... ah... it only supports MSI so it's forcing all the slots to use that? Jun 27 17:18:45 not sure Jun 27 17:19:08 but those 6 gigE interface are definitely the once addressed by those patches Jun 27 17:19:26 sorry, other way around... MFD's typically require MSI-X, the newer spec (with more interrupts but also more overhead). Jun 27 17:19:48 so what's the problem with the i82574L's? Jun 27 17:19:55 are they locking up under heavy load? Jun 27 17:20:07 theres two nics on irq 17 also Jun 27 17:20:53 3 and 7? yeah... Jun 27 17:21:31 philipp64: all of them stop responding to network traffic at "random" times Jun 27 17:21:33 I'd disable whatever hardware you're not using to free up some interrupts. Jun 27 17:21:38 (we don't have traffic monitoring) Jun 27 17:21:56 the weird thing is that the situation resolves itself after some time Jun 27 17:22:04 once this was after ~5m Jun 27 17:22:12 yesterday evening however it took more than 30m Jun 27 17:22:33 yeah... there's a timer that fires that notices that the NIC is locked up and forces it to reset itself. then it's fine for a while... until it locks up again. Jun 27 17:22:42 I see Jun 27 17:22:43 do you have the patch baked into your image or not? Jun 27 17:22:49 let me check Jun 27 17:23:19 if you do and you're still seeing the bug, then that's *really* weird and I'd contact Benjamin directly. Jun 27 17:23:34 I have the patch Jun 27 17:25:18 and weird messages appearing in dmesg? Jun 27 17:25:35 nothing at all Jun 27 17:29:05 what does your /etc/modules.d/e1000e file look like? Jun 27 17:31:42 What channel is for ordinary user chatter for lede / openwrt? Jun 27 17:32:30 profmac: #openwrt Jun 27 17:34:22 ooh Jun 27 17:39:00 duh. Thanks. Jun 27 17:41:45 :) Jun 27 17:42:22 I'm porting and FPGA board to the zynq target, Digilent Zybo Z7 (Zynq: Cortex A9 + FPGA) Jun 27 17:42:25 The board has two variants, "Zybo Z7-10" and "Zybo Z7-20" Jun 27 17:42:31 The differences between them are the SoC (XC7Z010 vs XC7Z020) and some components (RGB LEDs, HDMI CEC support, GPIO port) Jun 27 17:42:41 Are those differences enough to create two OpenWRT devices (boards/images), one for each variant? Jun 27 17:43:03 I need to know the current variant at runtime so I can program the FPGA with the correct bitstream Jun 27 17:43:04 and I haven't found a way to do it without two OpenWRT devices (with different device-tree compatible strings) Jun 27 17:46:21 would like to get an up/down vote on PR #6311 one of these days... tried to do the same thing with PR's to base-files and openssh but was told to do it as a virtual package standalone, so I did... Jun 27 17:46:42 luaraneda: having two "images" with two different DTSes sounds sensible to me Jun 27 17:47:08 luaraneda: we added distinct images for less reasons Jun 27 17:48:05 jow: Thanks. Upstream Linux has only one device-tree for the board, but I can send patches to split it into two variants and see if it get accepted Jun 27 17:48:25 jow: I'll send the patches to the kernel first, to get some feedback from them Jun 27 17:48:54 so apart from testing the dts compatible strings there is no way to differentiate the hw at runtime? Jun 27 17:49:08 if so then having distinct dts-es is fine Jun 27 17:49:45 if upstream rejects that and you want to stay close to it then defining two different devices in openwrt with different default packages might be a solution Jun 27 17:51:05 I could only find "/sys/devices/soc0/soc_id", without proper documentation Jun 27 17:51:19 with "bitstream" I assume you mean some kind of fpga firmware? If so, then having a firmware-xc7z010 vs. firmware-xc7z020 as packages might be a solution Jun 27 17:51:59 yes, is the FPGA firmware Jun 27 17:52:38 ... then have different Device/* sections selecting one of the two firmware-xc7z0[12]0 packages in DEFAULT_PACKAGES:=... Jun 27 17:53:43 Yes, I can create multiple packages, but my the idea is to be able to differentiate the variant at runtime from a generic package Jun 27 17:54:06 I see, well then having two images with two dts-es would be best Jun 27 17:54:24 yep Jun 27 17:55:04 Thanks for your help :) . I needed an extra opinion Jun 27 18:48:24 sent patches, probably did it wrong :D Jun 27 18:54:39 this is not a new problem on this particular (SLOOOW) hardware. Getting daemon.err uhttpd[21315]: sh: write error: Broken pipe in syslog. hints on basic debugging? Jun 27 19:06:58 jwh: both your netifd patches have no SoB Jun 27 19:07:43 i read that as 'son of a ...' but then I realised that had got to be wrong. Jun 27 19:20:03 Can Xiaomi WiFi 3G do 100 Mbps Shadowsocks? Jun 27 19:20:20 or Wireguard. Jun 27 19:20:26 chacha20 basically. Jun 27 19:20:32 dedeckeh: yeah, wasn't sure about that Jun 27 19:23:40 jwh:also prefix your patch with [PATCH 1/2 netifd]:system-linux.c: .... Jun 27 19:24:58 makes it easier to recognize to link the patches to a package Jun 27 19:25:18 mmm, how do I do that with git send-email? Jun 27 19:26:18 just make the patches with git format-patch and change the patch header Jun 27 19:26:23 ah Jun 27 19:26:29 ok Jun 27 19:26:34 then use git send-email to send them Jun 27 19:30:06 Can MediaTek MT7621 do 100 Mbps WireGuard? Jun 27 19:32:01 jwh: ldir taught me this: Jun 27 19:32:19 for e.g. a set of three patches: Jun 27 19:32:20 * ldir hides Jun 27 19:32:40 $ git send-email -3 --cover-letter --annotate Jun 27 19:32:50 that will combine three patches with an 'introductory' e-mail Jun 27 19:32:57 and give you the [PATCH 1/3] stuff Jun 27 19:33:11 ldir: yeah, you're my disclaimer ^_^ Jun 27 19:33:13 ye Jun 27 19:33:14 oh neat :o Jun 27 19:33:18 just in case he kills a kitten Jun 27 19:33:55 Very helpful page https://www.freedesktop.org/wiki/Software/PulseAudio/HowToUseGitSendEmail/ Jun 27 19:34:47 heh, I normally just use github/gitlab because I'm ultra lazy Jun 27 19:35:34 mm, who should I put for sob? myself? Jun 27 19:35:42 jwh: yeah Jun 27 19:35:47 rgr Jun 27 19:35:54 bleep Jun 27 19:36:01 jwh: you can configure git so you can just use git send-email $patch in the future Jun 27 19:36:04 it's pretty easy Jun 27 19:36:16 git config --global user.email "bla@bla.com" e.g. sets your from address Jun 27 19:36:20 yeah Jun 27 19:36:25 I've got the usual config, smtp etc Jun 27 19:36:30 you can set everything and the kitchensink Jun 27 19:36:31 ok Jun 27 19:36:34 I didn't use --annotate Jun 27 19:36:35 so you already know that :) Jun 27 19:36:37 probably should Jun 27 19:36:44 me neither, until ldir enlightened me Jun 27 19:36:53 don't forget the '-v2' switch for version 2 Jun 27 19:36:57 ^^ Jun 27 19:36:58 see Jun 27 19:37:04 the gift that keeps on giving :D Jun 27 19:37:07 lol Jun 27 19:37:09 ok Jun 27 19:37:17 need to get some food first though Jun 27 19:38:04 trust me on this....anything you think I know is because I fell into exactly the same trap before you did :-) Jun 27 19:38:20 * ldir goes back to strace Jun 27 19:39:13 :D Jun 27 19:39:56 that's how it works Jun 27 19:40:11 you spend countless hours on stuff, then you can dispense wisdom like a jedi Jun 27 19:43:11 I remember my first attempts at sending in a patch (dnsmasq version bumps as per usual) - oh my god was it painful - I threw my toys out of my pram in frustration a couple of times...as I'll admit I do on occasion. Trying to remember basic unix, learn git, learn openwrt build system, quilt (!!!!) was a stupidly steep learning curve. Jun 27 19:43:35 ugh, sod quilt Jun 27 19:43:42 makes me see red Jun 27 19:44:01 If I had known https://www.freedesktop.org/wiki/Software/PulseAudio/HowToUseGitSendEmail and https://raphaelhertzog.com/2012/08/08/how-to-use-quilt-to-manage-patches-in-debian-packages/ back then, well :-) Jun 27 19:45:47 quilt :( Jun 27 19:45:53 ldir: you had to pick up unix as well? Jun 27 19:46:43 anyway... i'm out guys, goodnight Jun 27 19:47:08 in previous life was a unix admin in the late 80's early 90's Jun 27 19:47:18 neat. Jun 27 19:47:40 i was literally throwing my toys out of my pram in the 80s ;) Jun 27 19:47:47 Pyramid OSx Jun 27 19:48:13 on which I learned (to a sort of fashion) the one true editor 'vi' :-) Jun 27 19:48:25 that came back like muscle memory. Jun 27 19:48:46 hehe Jun 27 19:48:53 that never goes away, muscle memory. Jun 27 19:49:09 i'm out though... lots of driving tomorrow, gotta be awake :P Jun 27 19:49:20 sleep well = drive safe (ish) Jun 27 19:49:31 ty Jun 27 20:16:21 How fast MediaTek MT7621 is? Jun 27 20:18:02 Can it run ChaCha20 at 100Mbps? Jun 27 20:21:13 buy one and see? Jun 27 20:21:13 :D Jun 27 20:23:20 Don't want to waste money. Jun 27 20:23:52 Maybe someone with this SoC can run shadowsocks bench on their router? Jun 27 20:23:59 using iperf3. Jun 27 20:24:29 bah, fucked it Jun 27 20:24:32 fml Jun 27 20:25:30 I really suck at this Jun 27 20:36:06 jwh: Signed-off-by: is still missing in your v2 Jun 27 20:37:20 I know Jun 27 20:37:22 :( Jun 27 20:37:29 I did it wrong Jun 27 21:11:28 ldir: lol Jun 27 21:13:09 he he - well it's what's in my mind so.. might as well say it :-) Jun 27 21:13:18 sure Jun 27 21:14:54 there, I think I did it right this time :D Jun 27 21:15:09 must remember to use --dry-run and maybe --confirm next time :D Jun 27 21:20:21 lol --dry-run became my friend last night after my first unaccompanied commit to the main repo. Jun 27 21:25:45 I managed to commit to the github mirror instead of the openwrt main repo...fortunately I realised my mistake and it was quiet (commit wise) so I was able to push the commit to where it should have gone...all the hashes matched so the mirror was fine. I have of course updated my git config to prevent such cockups in the future.....and --dry-run. Jun 27 21:27:06 heh Jun 27 21:27:10 "oops" Jun 27 21:27:56 actually there was a more colourful word involved, followed by a mild panic and thinking of 'how can I fix this?' :-) Jun 27 21:28:08 :D Jun 27 21:28:18 got the heart racing - needed a lie down after the fix Jun 27 21:28:49 nice cold beer Jun 27 21:30:25 I was already pretty merry...been up for 20 hours after 3ish hours sleep. I had a wireguard bump which was nice, simple, straightforward... I thought "how hard can it be?" Jun 27 21:30:52 heh Jun 27 21:31:07 don't commit while drinking Jun 27 21:31:49 turns out... harder that I thought.... and I'm a knob :-) Jun 27 21:32:29 there was no alcohol involved...just stupid from lack of sleep. Jun 27 21:33:17 talking of which - I'm turning in - g'night folks Jun 27 21:33:26 later Jun 27 21:44:01 ldir: chill Jun 27 22:48:37 Tapper: my money's on the switch driver being slow Jun 27 22:49:12 For the 3200x? Jun 27 22:50:07 yeah Jun 27 22:51:14 I can probably whip up a patch if you're interested. will require clearing settings Jun 27 22:52:05 I have the wrt3200acm and don't know how to run the same test that the bloke from the bugreport ran. Jun 27 22:52:18 neither do i :) Jun 27 22:52:30 That is why I was asking about how he did the testing befor. Jun 27 22:53:19 mangix if you do a patch I can try a build on my wrt3200acm to see if any thing breaks but... Jun 27 22:53:55 Or we can ask the dude to test it. By posting to FS Jun 27 22:59:03 Some more info: Jun 27 22:59:05 https://bugs.openwrt.org/index.php?do=details&task_id=1612 Jun 27 23:03:10 mangix he seems to think it is the router combo from his ISP Jun 27 23:04:18 The me and him have has one of the intel chipsets with the high pings under load, but they said that it was fixt in a fermwair update about 4 -5 monthes ago. Jun 27 23:05:03 Tapper: here you go anyway: https://gist.github.com/neheb/325f44a8e71fa6fe9250f0dbbfadb3f1 Jun 27 23:05:32 totally untested Jun 27 23:05:56 if it works, should be less problematic than the swconfig driver Jun 27 23:08:04 requires you to wipe /etc/config/network on sysupgrade Jun 27 23:08:07 or wipe all settings Jun 27 23:11:11 I need to install uduntu agin on windows lol I for got that I removed it to get windows 10 to update because of my small ssd. Jun 27 23:11:16 hahahah Jun 27 23:11:37 I will give it ago in the morning Jun 27 23:12:13 I need to grab me a new ssd now the prices are droping agin anyway Jun 27 23:12:49 prices for 3 TB HDDs aren't too bad at the moment Jun 27 23:15:27 sorry I will never go back to spinning rust! Jun 27 23:15:33 :-) Jun 27 23:15:40 just to slow Jun 27 23:15:51 Ok for storage but not for a mane drive Jun 27 23:16:12 I don't think I'll be able to afford replacing all my spinning rust with SSDs within the next 5 years Jun 27 23:16:22 :-) Jun 27 23:16:52 sure, the OS itself if on a SSD - but for data, VMs, etc. spinning rust is good enough Jun 27 23:17:02 mind you my hole PC needs a upgrade still running a core i7 920 Jun 27 23:17:43 the bitch gets hot in this nice weather we are having here in the UK Jun 27 23:18:48 Rite I am off to bed piece out people! Jun 27 23:19:04 mmm Jun 27 23:19:19 why is the gccgo stuff marked broken? Jun 27 23:26:01 mangix: do you want push those ipq8065 patches forward? I've been using the 4.9 variant for quite a while and they were fine. Jun 27 23:29:31 pkgadd: no Jun 27 23:29:48 ehhhhhh Jun 27 23:29:55 the rpm one could be merged Jun 27 23:30:09 the useless patches one partially Jun 27 23:31:04 the l2 scaling was NAK'd. I don't have enough kernel knowledge to fix it up Jun 27 23:33:09 I'm slightly afraid to add more patches to my patch queue that won't have a chance of being merged reasonably soon ;) Jun 27 23:43:24 i have a ton of those Jun 27 23:43:44 specifically the mt7621 ones Jun 27 23:45:11 I'm happy to have finally gotten rid of the worst ones (only stepping up ar71xx to kernel 4.14 is a continuing headache, two devices still needed it instead of ath79, but fortunately those aren't in daily use anymore) Jun 27 23:56:33 * mangix will play with ath79 soon Jun 27 23:56:45 time to port the ag71xx patches Jun 27 23:58:59 pkgadd: Can you tell me why I should avoid getting a router from US? Jun 27 23:59:06 You said something about regdom. Jun 28 00:01:58 koops: because the don't allow channel 12/13 and because their usage of the 5 GHz is exactly reverse to the allowed frequencies in europe (and I think .ru is similar to ETSI here) Jun 28 00:02:26 Can I somehow modify ART tables on a US router lol? Jun 28 00:03:22 yes, no, perhaps, depends on the driver - and device. but with a matching one you don't even have to try Jun 28 00:08:11 time to find old qca code Jun 28 00:08:15 this will be a nightmare Jun 28 00:08:47 pkgadd: Yeah, we use EU versions of routers. Jun 28 00:08:53 oh found it Jun 28 00:08:54 yay Jun 28 00:13:07 How fast is mt7621? Jun 28 00:13:31 Can it push 100Mbit/s WireGuard? Jun 28 00:20:27 probably not, among your list, mt7621 has the weakest CPU by a long margin Jun 28 00:21:35 ugh Jun 28 00:21:47 i hate undocumented qca code Jun 28 00:31:24 pkgadd: I found R7800 new for 126,88 EUR. Get it? Jun 28 00:40:32 hm, so gccgo builds just fine Jun 28 00:52:38 koops: I think I've seen new offers (shortly) down to about 140 EUR, other than that, the r7800 is a decent choice Jun 28 00:52:58 pkgadd: for Linksys? Jun 28 00:53:20 hmm? Jun 28 00:56:05 so whats the state of lantiq these days? Jun 28 00:56:29 doesn't seem to be any new devices :( Jun 28 00:57:15 the new devices are grx350 and grx500, nice hardware, not really supported in OpenWrt yet (there seems to be partial SOC support) Jun 28 00:57:44 ah Jun 28 00:58:33 would kinda like a small dsl modem, dm200 is ok but meh Jun 28 00:59:00 could do with ge port for baby jumbos Jun 28 01:02:33 iirc, you're from the US? I think ISP branded lantiq devices (which would be cheap on ebay) weren't that common there, so the dm200 looks like a nice option Jun 28 01:03:23 UK Jun 28 01:04:03 well, I go everywhere, but target market for these devices is the UK Jun 28 01:04:22 in that case, go with a BT Home Hub 5 type A, even if you'd degrade it to a pure modem (which would be slightly sad considering the the good wlan/ switch), there wouldn't really be anything to do wrong for ~8 GBP Jun 28 01:04:35 mm Jun 28 01:04:54 if I could get a decent number preflashed that would be pretty good I guess Jun 28 01:04:57 I wish I could use the HH5A over here in DeutscheGermanCountryLand. Jun 28 01:05:03 obviously the initial flashing is a bit involved (small test points) Jun 28 01:05:06 yeah Jun 28 01:05:10 but, alas, it doesn't fiddle annex b tunes. Jun 28 01:05:22 thats why I ruled it out before, theres no way the office guys will be able to manage it Jun 28 01:05:28 drmr: that's only an issue if you're on ADSL, VDSL is fine Jun 28 01:05:30 pkgadd: uhm, not that small. Jun 28 01:05:40 getting them to flash from edgeos is hard enouh Jun 28 01:05:43 enough Jun 28 01:05:54 in fact I distinctly remember molesting it with regular pitch pin headers. Jun 28 01:06:33 https://www.ebay.co.uk/itm/LEDE-OpenWRT-BT-Home-Hub-5-Unlocked-VDSL2-Gigabit-WiFi-Wireless-AC-VPN-router/352249464565 Jun 28 01:06:35 jwh: https://openwrt.ebilan.co.uk/viewtopic.php?f=14&t=965 that would be the easiest (and safest, as the via are easy to break while soldering) option, it needs about an hour to complete Jun 28 01:06:36 hm Jun 28 01:06:36 needs a temporary ground short to an smd resistor, but that can be done by just holding it there. Jun 28 01:07:00 but yes, preflashed devices are always an option - just more expensive Jun 28 01:07:07 yeah Jun 28 01:07:30 I mean, if I was in the office I'd just do it but I only visit every few months Jun 28 01:07:52 prepare- and put it in the mail Jun 28 01:08:41 after shipping it twice its probably going to cost the same as a preflashed one to be fair Jun 28 01:09:02 don't always have the ability to mail if I'm on the move anyway heh Jun 28 01:09:41 UK mail seems to be a tad on the expensive side, yes. I could send stuff from .de to the .uk for about half the price it cost me to get a bthub5 delivered to .de Jun 28 01:09:52 heh Jun 28 01:10:11 that seller is actually not far from the office Jun 28 01:12:42 if your really don't want wlan etc., the older ECI openreach modems (https://openwrt.org/toh/bt/vg3503j) could be another option Jun 28 01:12:45 -r Jun 28 01:13:14 mmm Jun 28 01:13:26 pretty much a dm200 though Jun 28 01:13:59 yes, just cheap(er) Jun 28 01:14:18 yeah Jun 28 01:14:39 * drmr actually has a speedport hs300 somewhere. good vdsl modems, no vectoring though. Jun 28 01:15:23 actually had an argument about that with one of our carriers a while back Jun 28 01:15:51 they were absolutely refusing to provide any support if we didn't use one of two models of modem Jun 28 01:16:09 (despite the dm200 actually meeting the required dsl spec) Jun 28 01:16:41 amusingly the HH5 isn't on that list heh Jun 28 01:16:43 understandable, though. Jun 28 01:16:50 well, kinda Jun 28 01:17:08 the spec isn't actually enforced and its just an approved/tested list Jun 28 01:17:20 imagine the load on their hotline if they had to phone support random stuff. Jun 28 01:17:22 so theres loads of devices that are fine but not on the list as the vendors haven't bothered Jun 28 01:18:12 well, wholesale supplier so kinda different Jun 28 01:18:42 and don't consider yourself as regular case. consider "my son in law left me this computer I use to connect to the wirelesses, and it is only blinking red. the model number is on the side, it says '230 volt'. how to fix please!" Jun 28 01:18:56 annoyingly the only half decent choice is broadcom Jun 28 01:19:51 heh Jun 28 01:20:55 its pretty silly really, since the carrier gains nothing by demanding their nonsense devices, no remote management or line state reporting etc Jun 28 01:21:33 whereas my openwrt builds for the dm200 as an example will regularly send me dsl stats so if the line starts to degrade it can at least be picked up before it dies entirely Jun 28 01:22:34 ah well, I just won't support circuits with black boxes on heh Jun 28 01:22:59 if it dies, have a spare ;) Jun 28 01:23:49 meh, they're opting for cheap circuits, they won't pay for a backup line (which probably follows exactly the same path anyway) Jun 28 01:32:09 vr200v looks kinda fancy Jun 28 01:32:25 antennas suck, but otherwise... Jun 28 01:42:17 drmr: Definitely no convenient headers in an HH5A. Jun 28 01:42:37 my memory may be wrong. Jun 28 01:42:37 And the 'test points' are really fragile vias. Jun 28 01:42:56 It is a long string of four letter words to work with :) Jun 28 01:44:13 I don't have any memory of it being super horrible, though. maybe my soldering is not as sucky as I think. Jun 28 01:44:45 I do BGAs and really, really small QFNs, I found it annoying. Jun 28 01:44:48 That should say something. :) Jun 28 01:45:46 Also I pulled a via out and had to fix that, which wasn't funny. Jun 28 01:48:48 I really suck at soldering, so I would probably lift every pad on the thing if I tried. xD Jun 28 01:51:40 tin foild, duct tape and -in the absense of fitting clamps- fingers to hold it all down worked surprisingly well Jun 28 01:52:13 but yes, it's tiny Jun 28 01:54:30 otoh, it doesn't have to be permanent. Jun 28 01:55:24 I found that small gauge wire is really essential for stuff like this. Jun 28 01:55:37 you don't need mains voltage cables running to the points. Jun 28 01:57:15 that reminds me, I have to pick up my stabilized 12V/ 200A PSU project... (aluminium rails instead of wires) Jun 28 01:57:36 200 *amps* ?! Jun 28 01:57:40 yes Jun 28 01:58:02 I think I remember using a three-pin header, soldering it down to some conveniently large ground point, and running thin wire to rx and tx. Jun 28 01:58:06 whats that for? Jun 28 01:58:19 lighting? Jun 28 01:58:22 ah Jun 28 01:58:45 lots of stuff, actually Jun 28 01:59:11 I need to revisit switched 12V at some point Jun 28 01:59:36 so I don't need to waste expensive pdu outlets on DC devices Jun 28 01:59:43 6 rectifiers, 8+1 transistors, a steel clothesline as resistors... Jun 28 02:00:04 I've been contemplating replacing a bunch of 12v wall warts with a proper PSU, and using splitter cables. but, eh, I think I can't be bothered. Jun 28 02:00:14 heh Jun 28 02:00:20 I did exactly that for some stuff Jun 28 02:00:26 just got one of those CCTV things off amazon Jun 28 02:00:38 and two huge heatsinks made from aluminium scrap Jun 28 02:01:21 jwh: I found that most if not all my devices use 5.5x2.5mm barrel plugs, most of the CCTV splitters are 2.1mm. :/ Jun 28 02:02:34 mm Jun 28 02:02:47 I haven't found a device the one I got hasn't worked with yet Jun 28 02:03:21 I'm not sure, maybe the 2.1mm ones are ... universal and fit 2.5mm too? Jun 28 02:03:42 I should just order some and try. Jun 28 02:04:31 I mean, the outer diameter is the same, and the inner one looks to be springy contacts on lots of plugs. Jun 28 02:04:43 eh, minor grievances. Jun 28 02:04:48 just trying to find the one I got Jun 28 02:05:30 https://www.amazon.co.uk/gp/product/B00C96U6WK Jun 28 02:05:37 honestly, the biggest problem I have right now is labeling my NAS drives* so I don't get them mixed up. :) Jun 28 02:05:53 *) recently converted to OpenWrt, to bring everything back on topic. Jun 28 02:06:03 I guess you could always just buy psu and the splitter seperately Jun 28 02:06:20 heh Jun 28 02:07:09 hmyeah, for some reason I don't particularly trust a €15 60W PSU as a single point of failure. but fair point, I'll just order the splitter cable, and if it works I'll supplement a nice, beefy PSU. Jun 28 02:07:09 I should really go to bed, got rb2011 testing to do Jun 28 02:07:30 tbf its been working pretty well, not heavily loaded though Jun 28 02:08:18 jwh: those are 2.5mm ones btw. :) Jun 28 02:08:28 got uh, couple of ubnt ERLs, dm200, on it Jun 28 02:08:29 yeah Jun 28 02:08:39 probably why I picked that one :D Jun 28 02:10:01 mh, I should call it a night, too. Jun 28 02:10:16 take care everyone. (waves) Jun 28 02:10:19 later Jun 28 02:10:47 remind me to try to find the openwrt contingent at the next toorcamp Jun 28 02:12:19 drmr: No, 5.5/2.1 plugs will not fit a 5.5/2.5 socket as the pin is in the socket Jun 28 02:12:28 5.5/2.5 plugs will badly fit 5.5/2.1 sockets Jun 28 02:56:06 pie: there were at least two of us over in the ROT503/PDXtech camp(s) **** ENDING LOGGING AT Thu Jun 28 03:00:02 2018