**** BEGIN LOGGING AT Mon Jun 25 03:00:03 2018 Jun 25 04:30:13 pkgadd: i really disagree with your opinion on Broadcom status Jun 25 04:30:59 pkgadd: I think it's support status is great if you don't care about WiFi Jun 25 04:31:24 pkgadd: almost everything is supported with upstream code & drivers Jun 25 04:32:12 pkgadd: look how few files there are in the target/linux/bcm53xx/patches-4.14/ and please remember that 0xx are backports from upstream kernel Jun 25 04:32:27 koops: if you don't care about WiFI, Broadcom platform is OK Jun 25 04:32:54 koops: the only unsupported hardware part you may care about is hardware flow acceleration Jun 25 04:33:17 koops: it's mainly because support for offloading was added so recently to the kernel Jun 25 04:33:24 it's definitely possible to support it Jun 25 04:33:29 (now) Jun 25 04:41:52 the worste part of sending patches upstream are the OF binding docs Jun 25 04:48:07 well, I got used to it and somehow learned how to write them Jun 25 04:48:10 it's not that bad now :) Jun 25 04:48:22 it just takes some time Jun 25 04:50:53 hi i have latency and packet loss issues . i am using archer c7 v2 openwrt and using a mifi usb tether to get internet . Jun 25 04:51:03 https://pastebin.com/8dkuURt8 Jun 25 04:51:22 https://pastebin.com/jC4nZGk5 Jun 25 04:52:31 first link is my dmesg, /etc/config/wireless /etc/config/network and ping Jun 25 04:52:38 2nd link has ifconfig Jun 25 04:55:46 disabled both firwalls Jun 25 05:13:52 sorry i posted here , i reposted it in #openwrt channel Jun 25 07:12:02 jow: ping - since luci 5142e40f9 the firewall port forwards & traffic rules pages don't show rule names (for me at least) Jun 25 07:12:30 ldir: should be fixed now Jun 25 07:12:53 oh have you pushed more in the past 10 mins? Jun 25 07:12:54 Hello all .. I am getting a strange error in my OpenWRT CC custom build firmware .. When i create or edit an SSID , i am getting "Bad Gateway: The process did not produce any response" ... Jun 25 07:13:05 ldir: yes Jun 25 07:13:32 Any help?? jow ?? anyone ? Jun 25 07:13:35 jow: dammit! you're just too quick for me :-) Jun 25 07:15:00 jow: also the realtime graphs appear b0rked... did you know? Is that similarly fixed in past 10 mins? Jun 25 07:15:14 whats borken about them? Jun 25 07:15:22 no data Jun 25 07:16:47 don't see luci-bwc running Jun 25 07:20:21 Obaid: no idea, CC is getting on a bit now. Jun 25 07:21:12 ldir: It is a very strange error .. :( Jun 25 07:21:23 ldir: fixed too Jun 25 07:21:45 Obaid: well it just means that luci either hangs or crashed Jun 25 07:21:53 maybe libiwinfo is missing Jun 25 07:22:21 jow: There is a libiwinfo in /usr/lib .. Jun 25 07:22:29 libiwinfo-lua Jun 25 07:24:36 * ldir pulls & builds Jun 25 07:25:02 jow: Ok, i don't see this file in either /lib/ or /usr/lib .. :/ Jun 25 07:25:15 jow: from where i can get this file ?? Jun 25 07:26:49 should be in /usr/lib/lua/iwinfo.so Jun 25 07:26:56 and be part of libiwinfo-lua Jun 25 07:28:32 jow: yes, there is iwinfo.so in /usr/lib/lua .. but i don't see libiwinfo-lua in my AP.. Jun 25 07:29:03 Obaid: then this is not the problem Jun 25 07:29:46 jow: yes .. so what could be the possible cause of it ?? this error keeps coming and it's getting on my head now Jun 25 07:30:28 no idea Jun 25 07:30:28 jow: I tried rebooting several times .. Jun 25 07:30:39 I guess luci either crashes or times out Jun 25 07:31:27 jow: Any place in AP where i can find the logs ? logread/dmesg aren't providing some logs for this error .. Jun 25 07:31:51 there is only logread or dmesg Jun 25 07:31:59 does it happen with an official build too? Jun 25 07:32:30 jow: you mean the pre-compiled build ? Jun 25 07:32:37 yes Jun 25 07:33:18 jow: well, i haven't checked those .. I don't think official build will have such problem Jun 25 07:41:14 ok realtime graphs fixed. Am still missing rule names in firewall -> traffic rules. What have I done wrong? Jun 25 07:43:48 ldir: seems to be some bug in the bootstrap theme Jun 25 07:45:11 ok, what's the preferred theme? :-) Jun 25 07:46:59 as a workaround you can use the openwrt one, will fix bootstrap now Jun 25 07:53:16 ldir: should be fixed with bootstrap too, now Jun 25 07:55:32 great - will test. I stick with bootstrap 'cos it's the default. Jun 25 08:07:34 jow: better, not 'perfect' - still missing if no rules. Also no column heading/title. Jun 25 08:39:20 ldir: no heading for the name column is intentional Jun 25 08:39:36 not sure what you mean with "missing if no rules" Jun 25 08:41:44 the rule name column is missing if there are no rules to display. I'm guessing because of the intentional 'no heading' Jun 25 08:42:41 checking... Jun 25 08:42:46 is it possible that the preint script fails in case it can't write to /dev/kmsg? Jun 25 08:42:53 So without any rules configured I see "Match. Action. Enabled" columns. Jun 25 08:43:18 if any rules defined I then see " (space). Match. Action. Enabled" columns Jun 25 08:43:49 dwmw2_gone: ping Jun 25 09:03:25 * ldir switches to the openwrt theme Jun 25 09:07:04 dwmw2_gone: unping Jun 25 09:08:06 * ldir wonders what blogic is breaking? :-) Jun 25 09:12:40 ldir: should be fixed too, now Jun 25 09:13:20 * f00b4r0 reads about TLBleed - side channel attacks are getting cool Jun 25 09:14:22 blogic: if any use have had https://github.com/openwrt/openwrt/pull/1073 ath79 for the archer c7 v2 - been running the AP with that for 48 hrs & took a deep breath with my main router last night... running 12hrs+ nothing horrible noticed. Jun 25 09:15:51 jow: great - will give that a go shortly :-) Jun 25 09:15:52 was blaming ath10k for latency and outright dropped packets. turned out it's this damn macbook, and may even be a hardware issue, since windows has similar problems Jun 25 09:16:11 could've used my weekend in better ways.. Jun 25 09:17:39 mm, has anyone had a look at the ER-6P yet? looks like its straight cpu ports rather than a switch, might be nice Jun 25 09:18:41 jwh: why might it be nice? because port<>port traffic needs to be forwarded in software now? Jun 25 09:18:48 or is it simply a dsa switch? Jun 25 09:18:54 no DSA, real cpu ports Jun 25 09:18:59 meh Jun 25 09:19:14 I actually have a fair few cases where I *don't* want a switch in the device Jun 25 09:19:27 especially if theres no DSA Jun 25 09:19:59 things like actually forwarding traffic between 2 ports without hairpinning to the slow cpu port and back Jun 25 09:20:13 ldir: can't you merge it on your own ;-) Jun 25 09:20:47 lack of switch also makes it easier to support since they utterly refuse to give up the details for their hardware Jun 25 09:21:22 will see if I can get one ordered for my openwrt lab :D Jun 25 09:21:34 oh wait Jun 25 09:21:42 mkresin: what my highly complicated luci PR? ;-) Jun 25 09:21:55 wikidev claims theres a switch Jun 25 09:22:27 ldir: no, the ath79 PR you mentioned Jun 25 09:22:30 jwh: why support the apple of networking with money :/ Jun 25 09:22:38 umm Jun 25 09:22:46 because they make devices with useful socs? :D Jun 25 09:23:04 can't argue that :D Jun 25 09:23:17 fraction of the cost of other OEMs using the same soc Jun 25 09:23:19 heh Jun 25 09:24:05 QCA8337N Jun 25 09:24:07 fun times Jun 25 09:24:38 also, their products look good. Important point :3 Jun 25 09:24:44 heh Jun 25 09:28:12 mkresin: oh, err, it seems not yet. Jun 25 09:29:54 ldir: you need to provide jow with your ssh key Jun 25 09:30:08 and then you need to follow the howto for adding your key to the keyring Jun 25 09:30:12 * ldir suspects he needs to go through some initiation....and Madagascan fruit tree bats are hard to come by. Jun 25 09:30:28 Morning all, I've been discussing an issue with running openwrt on a Linksys WRT32X on the the lede-project forums. It was suggested I log it as a bug but the bug logging page says to make sure it *is* a bug before logging. ANyone got five minutes for me to run it by them before I log it? Jun 25 09:30:44 ldir: no, just send a ssh key Jun 25 09:30:55 ldir: sorry ... Jun 25 09:33:33 oKey doKey - will do. Sorry, bad puns are not optional :-) Jun 25 09:33:42 figure Jun 25 09:34:14 there goes your chance to make ldir do something silly Jun 25 09:34:15 darn Jun 25 09:36:38 ok so, there isn't a switch after all, might be worthwhile investigating Jun 25 09:36:52 I shall obtain one Jun 25 09:37:44 ccgjp: dont ask to ask Jun 25 09:38:12 Okay, sorry, first time on channel, wasn't sure of etiquete! :-D Jun 25 09:40:31 Right, high level summary: I'm seeing very poor ping times from the internet to the WAN interface using IPv4 on a Linksys WRT32X running davidc502's build 7210, recent 18.06 snapshot and the stock Linksys firmware (which is, I believe, Designated Driver under the linksys web frontend... Jun 25 09:41:43 With my old Apple Airport Extreme connected to my cable modem I get a very steady 20ms ping time but moving to the WRT32X I see very bad jitter, with ping times typically fluctuating between 20ms and 40ms. Jun 25 09:42:09 what does stock 18.06 say ? Jun 25 09:42:15 is davidc's build a recent 18.06 snapshot? what device is running the stock linksys firmware? Jun 25 09:42:20 without the david502 patches Jun 25 09:42:37 Here's a smokeping chart. The plots to the left are the WRT32X on 18.06, to the right is the Airport Extreme: https://www.osx.ninja/stock_to_airport.jpg Jun 25 09:43:06 blogic: 18.06 snapshot behaves exactly the same a the davidc502 build Jun 25 09:43:43 karlp: It's the same WRT32X that I"ve been running the stock, davidc502 and 18.06 snapshot on Jun 25 09:44:11 NB these ping times are with no to very little load on the WAN link Jun 25 09:44:41 no idea, sorry Jun 25 09:44:53 and what does stock linksys firmware do? Jun 25 09:45:36 karlp: stock linksys firmware behaves the same way, much higher lateny and jitter than the old Airport Jun 25 09:46:36 I'm also going to be raising this on the linksys support forums later today Jun 25 09:48:24 That forum thread for the davidc502 builds suggested this was reminiscent of issues wuth the R7800 so urged me to log it as bug Jun 25 09:50:34 FYI, specific snapshot I was testing was 7029 as there wasn't an RC1 buld for WRT32X at that time Jun 25 09:50:57 sorry *7079*, not 7029 Jun 25 09:54:49 https://store.ubnt.com/products/edgerouter-12 Jun 25 09:54:50 cue Jun 25 09:54:51 cute Jun 25 09:55:11 195 dollars beta (need to be in the US and be logged in with beta enabled account to see it) Jun 25 09:55:29 10 swithed copper ports, 2 sfp ports Jun 25 09:56:00 switched* Jun 25 09:56:46 https://cdn.shopify.com/s/files/1/1439/1668/products/ER-12_Top_Angle_1024x1024.png?v=1527284478 Jun 25 09:58:02 can't even access the link dude. Jun 25 09:58:10 oh? Jun 25 09:58:10 the internet of the usa Jun 25 09:58:16 oh the product page Jun 25 09:58:17 yeah Jun 25 09:58:21 trumpistan Jun 25 09:58:28 should be able to see the img though Jun 25 09:58:40 blogic: gdpr-istan Jun 25 09:58:48 lols Jun 25 09:58:53 * blogic puts on the bonanza music Jun 25 09:58:55 yehaaa Jun 25 09:59:22 but we also got upload filters now Jun 25 09:59:33 so its like the chinese gfw, just cooler, because it goes reverse! Jun 25 10:00:56 lulz Jun 25 10:01:13 meh, sold out anyway Jun 25 10:01:14 restricted to ubnt-beta peoples, too Jun 25 10:01:14 lamers Jun 25 10:01:19 gimme Jun 25 10:01:25 at 195$ its hard to say no Jun 25 10:01:33 easy to say no Jun 25 10:01:34 even if it never works, can sell it at retail cost later :D Jun 25 10:01:42 probably 500-600$ Jun 25 10:02:12 blogic: Rawhide! Jun 25 10:02:53 ldir: :-D Jun 25 10:03:17 the EdgeRouter 12 delivers up to 3.4 million pps for 64-byte packets. Ten Gigabit Ethernet ports offer copper connectivity, while two SFP ports offer fiber connectivity. Jun 25 10:03:21 dowant Jun 25 10:03:46 sounds like octeon-tx Jun 25 10:03:46 it looks like they're all switched ports, wonder what chip they're using Jun 25 10:03:56 marvell most likely Jun 25 10:03:56 its uh, octeon III by the looks of it Jun 25 10:04:08 same as the 4p/6p Jun 25 10:04:27 just inspecting the firmware Jun 25 10:04:41 bleh, looks like QCA Jun 25 10:04:46 marvell would be an improvement Jun 25 10:06:06 the print on the front of that image suggests only the first bunch of ports are really switched, so probably just their go-to 8337N Jun 25 10:06:49 or maybe one of the new 85xx ones Jun 25 10:09:21 jow: ping - pm/query? Jun 25 10:09:28 sure Jun 25 10:24:45 meh, can't get binary growth under 642 bytes Jun 25 10:24:54 my C kung fu is not strong enough Jun 25 10:26:10 can I get squashfs to give me the sizes of a binary its packed? Jun 25 10:26:20 checking size of the staging dir may nto be that useful Jun 25 10:26:23 not* Jun 25 10:41:33 . Jun 25 11:32:16 jow: Is possible to merge https://forum.lede-project.org/t/pc-engines-apu2-leds-not-working-in-lede-17-01/12722/11 (git format) to openwrt-18.06 ? Jun 25 11:32:35 nbd: ping Jun 25 11:32:39 it applies also for trunk Jun 25 11:33:36 muhaha: needs to be a git-am'able patch with signed-off + commit message Jun 25 11:33:39 or a PR Jun 25 11:34:57 jow: will you do cherry pick to openwrt-18.06 (i guess there are not diffs) if I submit this patch ? Jun 25 11:35:05 maybe Jun 25 11:38:51 I think there's a low-level bug in ustream Jun 25 11:39:23 hmmm Jun 25 11:39:24 for a while I noticed that various uhttpd resources fail to properly load or are delivered truncated Jun 25 11:39:31 when stracing the server process I see Jun 25 11:39:44 I can't figure out the most sensible place to put some boot stuff, init.d seems too late Jun 25 11:40:01 all its doing is checking if overlay is there, if not doing sysupgrade Jun 25 11:40:19 although I guess it doesn't really matter as long as it happens Jun 25 11:40:30 https://pastebin.com/Tq3yVbHm Jun 25 11:41:08 partial write(), followed by an attempt to write() the rest (interrupted with EAGAIN), followed by droppign the descriptor Jun 25 11:51:48 ok another question Jun 25 11:52:10 can I easily include files in the image for *just* initramfs? Jun 25 12:20:49 jow: Is it enough ? https://pastebin.com/TsLtpMph Jun 25 12:22:38 jow: can confirm the " "Name. Match. Action. Enabled" columns thing is fixed. Thank you Jun 25 13:17:00 jow: Where does the config.seed for the release and snapshot builds originate? Jun 25 13:20:19 the mountdirectory ist set to 755 afer boot. is it possible this set to 777 in /etcc/config/fstab? Jun 25 13:20:32 sorry for my crap endlish :) Jun 25 13:20:39 *english Jun 25 13:23:53 Monkeh: make diffconfig Jun 25 14:02:47 jow: Is it enough ? https://pastebin.com/TsLtpMph Jun 25 14:04:23 muhaha: partially Jun 25 14:04:41 the patch description is vague, no idea what this diff is finxing Jun 25 14:07:10 ajaj, in coreboot(pcengines apu firmware) was changed boardname from apu to PC engines apuX Jun 25 14:07:16 *name Jun 25 14:11:39 hi, is there a way to set individual mcs rates? (ath10k) Jun 25 14:13:57 sorry; chipset is Marvell Jun 25 14:27:24 guys Jun 25 14:27:43 quick fire, box with at least one copper and one sfp port that works Jun 25 14:28:29 someone made a boo boo and delivered a circuit with the wrong presentation :D Jun 25 14:34:15 hm Jun 25 14:34:17 Removing obsolete file /usr/share/domoticz/www/secpanel/js/less.js.gz. Jun 25 14:34:18 Configuring domoticz. Jun 25 14:34:18 Collected errors: Jun 25 14:34:18 * file_sha256sum_alloc: Failed to open file /etc/domoticz/plugins/examples/UDP: No such file or directory. Jun 25 14:34:19 * file_sha256sum_alloc: Failed to open file /Discovery.py: No such file or directory. Jun 25 14:34:24 root@LEDE:~# ls /etc/domoticz/plugins/examples/ Jun 25 14:34:25 BaseTemplate.py Dlink DSP-W215.py HTTP.py MQTT Publish.py Pinger.py UDP Discovery.py Jun 25 14:34:31 jow is it possible that the /sbin/init procress completely fails if access to /dev/kmsg is not permitted? Jun 25 14:34:33 is that an opkg bug? Jun 25 14:34:33 orrrrr, does the sfp port on the rb2011 work now? Jun 25 14:35:17 dwmw2_gone: looks like it doesn't like spaces in filenames ? Jun 25 14:35:26 dwmw2_gone: anyway, examples should be dropped from the package Jun 25 14:35:32 no ? Jun 25 14:35:35 yeah Jun 25 14:36:15 stintel: I'm going to commit https://git.openwrt.org/0baa354b32ca7fc1445e1efb21f79c7c34084200 later Jun 25 14:36:50 stintel: should make the led config of the ath10k-phy0 led obsolete Jun 25 14:37:54 mkresin: you'd still need the CONFIG_ATH10K_LEDS afaict Jun 25 14:38:42 mkresin: feel free to grab that patch from my staging tree and drop the config led part from the commit message Jun 25 14:38:49 stintel: yes it is still required. but you can the drop the part about the led config from your commit message Jun 25 14:38:55 mkresin: oh :) Jun 25 14:38:59 LOL Jun 25 14:54:16 morning Jun 25 15:00:55 hm, what is QCA9557? Jun 25 15:00:58 is that ipq? Jun 25 15:08:52 https://gitlab.labs.nic.cz/turris/openwrt/commit/c62a189e9cbf5878e790d60ebee453312ff5f3f5 Jun 25 15:08:57 hmmmm Jun 25 15:14:33 hi, is there a way to set individual mcs rates? (iwlwifi) Jun 25 15:15:13 *mwlwifi Jun 25 15:17:26 hi is there here that have problems with iphones on nbd mt76 wireless? Jun 25 15:17:26 jwh: no, abgn wisoc Jun 25 15:17:48 aah Jun 25 15:34:20 mmm Jun 25 15:34:21 https://mikrotik.com/product/hex_s Jun 25 15:34:28 looks suspiciously like an mtk Jun 25 15:34:48 ah it is Jun 25 15:35:50 no console, sucky Jun 25 15:43:05 nice box. Looks like something I could use. Running owrt shouldn't be extra hard either Jun 25 15:47:45 yeah Jun 25 15:47:50 not gonna happen without a console though Jun 25 15:48:00 and mikrotik probably don't leave the uart headers on Jun 25 15:48:01 heh Jun 25 15:48:09 in my experience they do Jun 25 15:48:25 really? Jun 25 15:48:27 all the boards I ported had uart pins available Jun 25 15:48:37 boy that thing is cheap Jun 25 15:48:40 i'm very tempted Jun 25 15:48:42 handful of boxes I've checked (mostly the bigger stuff) doesn't have any if theres no console port Jun 25 15:48:52 while you're at it you can add sfp support :DDD Jun 25 15:49:05 er-x isn't much more btw Jun 25 15:49:08 but has 256M of nand Jun 25 15:49:30 and ralink uboot, rather than routerboot Jun 25 15:49:48 true Jun 25 15:50:03 I've still got mine wired up somewhere Jun 25 15:50:10 not got much use for them though Jun 25 15:50:18 not without the sfp port anyway heh Jun 25 15:51:35 no matter how much I read dts though I can't figure out what it does Jun 25 15:53:19 the er-x can't do proper poe-in though Jun 25 15:54:13 hmm. the hex-S is available on amazon Jun 25 15:54:27 trusay Jun 25 15:54:33 tempted to order one, just to take a peek at the pcb and send it back if I don't find the expect uart Jun 25 15:55:17 if its mostly a generic mtk which it looks like it might be Jun 25 15:55:27 may be able to boot existing on it and at least get ethernet Jun 25 15:55:32 also, who needs 256MB nand when you have SD :) Jun 25 15:55:50 my images are pretty busy ;D Jun 25 15:56:06 the sd card is a pretty cool feature Jun 25 15:56:10 yeah Jun 25 15:56:59 * jwh spots a 20km optic with a .1m patch lead Jun 25 15:57:05 hello toasty lasers Jun 25 15:57:39 I ordered an rb2011 anyway Jun 25 15:57:55 will be nice if it works Jun 25 15:58:08 istr it's supported Jun 25 15:58:20 yeah, sfp is my concern Jun 25 15:58:28 but it seems like the only one likely to work Jun 25 15:59:51 https://www.mikrotik.com/product/RB2011UiAS-RM Jun 25 15:59:53 tht one Jun 25 15:59:54 that* Jun 25 16:00:23 sfp is on shared cpu port but can't have everything Jun 25 16:02:35 aparcar: wouldn't surprise me Jun 25 16:16:44 what is so cool about sftp anyways? i've never used it :/ Jun 25 16:16:47 *sfp Jun 25 16:17:40 https://en.wikipedia.org/wiki/Small_form-factor_pluggable_transceiver Jun 25 16:18:51 yeah but those are usually 1gbit too. what's the appeal over copper? Jun 25 16:20:07 movi: distance Jun 25 16:20:26 presentation, scalability Jun 25 16:20:37 also Jun 25 16:20:40 so it's basically the "wohoo optical" port? ;) Jun 25 16:21:13 its woohoo my customer that is 30miles away can get 10G Jun 25 16:21:13 :P Jun 25 16:21:40 jwh: not on that toy you linked he can't :P Jun 25 16:21:46 then you add wdm, so I can have multiple links per pair Jun 25 16:21:59 or even without wdm, I can have bidi and have a link per strand Jun 25 16:22:05 well no, not on that Jun 25 16:22:15 ok i was wondering if i was missing out on something Jun 25 16:22:39 I don't need to repatch anything, can have physical link failover with mirrors too Jun 25 16:22:42 etc Jun 25 16:23:38 I really just want a nice compact box (er-x, that mikrotik hex type form factor) that has at least 1 copper and 1 sfp port Jun 25 16:23:42 just a gig will do Jun 25 16:24:02 customers with more tend to have higher requirements anyway so they just will not do Jun 25 16:27:53 but for those, x86 exists Jun 25 16:29:38 oh this is cute Jun 25 16:29:44 # ethtool -m eth2 Jun 25 16:29:44 Cannot get module EEPROM information: I/O error Jun 25 16:30:19 oic Jun 25 16:31:04 jow: hm Jun 25 16:31:49 devstatus says link-supported and advertising is 1000F Jun 25 16:31:59 but its actually 10G Jun 25 16:32:07 the "speed" value is right Jun 25 16:32:36 is that just missing definitions for >1G? Jun 25 16:32:50 possible Jun 25 16:33:09 "speed": "10000F", Jun 25 16:33:11 that looks fine Jun 25 16:33:54 might be worth adding, maybe also 2.5G and all that nonsense too, at least one supported platform (mvebu?) does 2.5 iirc Jun 25 16:35:06 huh, its just querying ethtool Jun 25 16:35:07 weird Jun 25 16:36:07 aaah Jun 25 16:36:29 ethtool_link_modes Jun 25 16:39:56 apparently ETHTOOL_GLINKSETTINGS is the hip new way of doing it anyway Jun 25 16:44:19 we should create an #openwrt-jwh channel Jun 25 16:44:54 :( Jun 25 16:44:54 hey at least someone's talking in #openwrt-devel :P Jun 25 16:45:05 I just want everyone to have nice things! Jun 25 16:45:14 * Borromini lines up Jun 25 16:45:15 everyone deserves something nice Jun 25 16:45:21 get me that lambo please. Jun 25 16:45:24 lol Jun 25 16:46:42 if everyone had a hobby as awesome as us, there'd be less wars Jun 25 16:46:51 they are all just miserable and bored, .... Jun 25 16:47:15 i don't work in IT but my line of work also has burnouts i guess :P Jun 25 16:47:23 heh Jun 25 16:47:27 'hey i got this fixed.' Jun 25 16:47:34 'bah why is there another new release???' Jun 25 16:47:43 'gotta backport that stuff *again*' Jun 25 16:51:26 dwmw2_gone: wondering if we should create an openwrt-upstream ML that we can CC for our upstream work Jun 25 16:53:39 blogic: I'd vote yes - it's hard keeping track of all the work if you're not directly involved Jun 25 16:55:05 Monkeh: so my remote hacking setup plan is growin Jun 25 16:55:22 i ordered 8 esp based power sockets Jun 25 16:55:45 once they arrived i'll write some FW for them using the espressif SDK Jun 25 16:55:55 and then i'll ping you about that PCB :-D Jun 25 16:56:07 Mmm, cheapo relays.. :) Jun 25 16:56:16 9 euros/ each Jun 25 17:05:34 well, hip ethtool is just messy Jun 25 17:33:49 been an ath9k user for nearly a decade, got the interest in MTK wifi, very disappointed thus far, it's like I'm always getting subpar 11g performance, let alone 11n/11ac Jun 25 17:34:46 with mt7621a and mt7620a/mt7620n all tested within 1m/3m/10m line-of-sigh, the performance is so horrible why they are so popular? running 17.01.04 and 18.06.rc1 on them Jun 25 17:35:17 is the typo intentional? :P Jun 25 17:35:53 GL.inet 300M gave me 20mbps at best, zbt1326(max 1300mbps) gave me best around 40mbps in 2.4Ghz while 90mbps on 5Ghz(11ac) Jun 25 17:36:26 yo, time frame on a new release? Jun 25 17:37:05 based on RSSI measurement, GL.inet's two little antenna reacher far more than ZBT's 4 big ones, in short, ZBT's signal starts to degrade dramatically a few meters away Jun 25 17:37:36 Borromini: :) s/sigh/sight, here it means the same thing for me Jun 25 17:38:49 even measuring with 1-3m the performance is still terrible to say the best Jun 25 17:38:58 ausjke: i can't complain about mt76 5 GHz myself but I have not a heavy wireless use Jun 25 17:39:01 * user Jun 25 17:39:34 what's the speed you got then Jun 25 17:39:58 i should check. Jun 25 17:40:06 you're on vht80? Jun 25 17:40:08 i use iperf3 to benchmark with a laptop hooked to its LAN port as the iperf3 server Jun 25 17:40:14 yes vht80 Jun 25 17:41:24 just give me some real 11n speed i will be in heaven and don't really care how soccer plays out in the coming days Jun 25 17:42:19 now when i read c2600, c1300, c1750... i will dive them by 100 to stay realistic, at leat with MTK chips :( Jun 25 17:42:30 s/dive/divide/ Jun 25 18:10:04 ausjke: just copied some files over (audio) and i got up till 20 MBps Jun 25 18:10:29 I got bored of making it work, so I just did this instead... https://github.com/joeholden/netifd/commit/6c38041e9ded628b82ecb6769f85c6b6cc2548b7 Jun 25 18:10:42 ausjke: linux 4.16 client with Intel wireless 8260AC Jun 25 18:13:53 jow: https://pastebin.com/uhB7C4NL whats wrong with this again ? Thanks Jun 25 18:20:20 Borromini: nice at least you are in the 11n ball park Jun 25 18:20:38 * ausjke stays in the middle 11g for ages, literally Jun 25 18:21:31 wifi speed is over rated so much anyways across the board Jun 25 18:21:50 muhaha: the commit message doesn't tell us WHY we need the change. Sell it to us, show us how we would be batshit crazy not to have this code. Jun 25 18:22:18 ausjke: the newer the genreation, the more inflated the numbers are. Jun 25 18:22:19 excluding overhead of all kinds, at least give me half of the marketing claimed speed... Jun 25 18:22:34 I never got to 20 MBps with my 802.11an setup Jun 25 18:22:46 you won't even get to half i think with 802.11ac. Jun 25 18:23:41 muhaha: tell us how you tested it and on what hardware. Jun 25 18:24:05 if i'm a lawyer maybe i can make some cash by class action against the wifi vendors, ha Jun 25 18:25:02 you could... only lawyers get rich off that. Jun 25 18:25:08 i'm out, later guys Jun 25 18:26:02 * ldir also exits - crazy early start tomorrow, so early it barely is tomorrow. Jun 25 18:26:42 ldir: Huh, without this patch, on PC engines APU2C4 on coreboot 4.6.X does not work gpio-nct5105c .. Module is not loaded. Jun 25 18:27:56 That's EXACTLY the sort of thing that should go in the commit message. Jun 25 18:28:04 and I really have gone now Jun 25 18:33:54 blogic: do you know if we have ever tried to upstream target/linux/ramips/patches-4.14/0024-GPIO-add-named-gpio-exports.patch? Jun 25 18:46:04 neoraider: I think this was the last attempt from someone to get it upstream: https://www.spinics.net/lists/devicetree/msg08604.html Jun 25 18:51:58 Hmm, I see Jun 25 19:03:34 Help Jun 25 19:04:12 Wr841n v13 USB dts Jun 25 19:04:25 that's specific, but not veruy usefully specific :) Jun 25 19:04:41 Help Jun 25 19:05:12 Wr841n v13 USB mod help Jun 25 19:06:15 Tplink wr841n v13 usbmod dts Jun 25 19:06:57 ok then... that was easy Jun 25 19:17:00 karlp: looks like he mistook irc for google search ;P Jun 25 19:20:35 hey I'm a beginner trying to port the netgear wndr4300 from ar71xx to ath79. I have made some progress (https://github.com/Zsub/openwrt/tree/ath79-wndr4300) but I'm struggling to find the correct values to define the wifi in DTS. Is someone online who could give me some pointers? Jun 25 19:24:03 To me, it looks like the DTS for the TP-link tl-wdr4300 should be kind of similar, is that assumption even correct? Jun 25 19:28:38 if the request "If you have one of these and you're happy to crack it open and post PCB/SoC pictures and/or serial logs" is still wanted on https://openwrt.org/toh/arcadyan/ar7516 I have one here Jun 25 19:29:40 visual ID against the linked page checks out https://shop.ee.co.uk/broadband/ee-bright-box-router it was sent out by EE a couple months ago Jun 25 19:30:34 (I tried to test it as a replacement for my router but its web interface won't let me set a username that isn't @ an EE domain :/ ) Jun 25 19:31:05 jow, whats happened to the luci theme it resembles the basic version its awful Jun 25 19:31:59 even though I try to change themes nothing seems to change too much Jun 25 19:35:25 trine: browser cache? Jun 25 20:01:17 I need a travel router that can run Shadowsocks or Wireguard at least at 100 Mbit/s. It should have a USB port and at least one at least 100 Mbit/s LAN port. Jun 25 20:01:39 The encryption used is chacha20-poly1305 Jun 25 20:01:51 would you like a monkey butler too Jun 25 20:01:57 perhaps a moon on a stick sir? Jun 25 20:02:15 I need similar though Jun 25 20:02:24 lan port optional, need usb (or built in lte) and wifi Jun 25 20:02:29 small as possible Jun 25 20:02:45 also ultra low power for battery Jun 25 20:03:10 ChaCha20 is about 3x faster than AES. Jun 25 20:03:31 it also has exactly zero hardware acceleration support Jun 25 20:03:34 Yeah, USB power is fine. Jun 25 20:03:53 really, especially if it needs to be low power Jun 25 20:03:56 you want hardware crypto Jun 25 20:03:57 I disconnected, so I lost some messages. Jun 25 20:04:11 cpu is just a waste Jun 25 20:04:20 Not with ChaCha20. Jun 25 20:04:33 AES is faster with hardware acceleration though. Jun 25 20:04:42 but without it it's 3x slower. Jun 25 20:04:51 GCM is pretty fast Jun 25 20:05:01 thats why network protocols use it Jun 25 20:06:07 AES-GCM can't really be faster than AES-CTR Jun 25 20:06:44 GCM lends itself better to existing cpu design Jun 25 20:06:57 and also is relatively low complexity Jun 25 20:07:11 but who even uses CTR Jun 25 20:07:20 ssh Jun 25 20:07:31 I don't think i've actually seen it in the wild Jun 25 20:07:34 mt7621 cpu can do +100Mbit with wireguard. Jun 25 20:07:37 yeah but ssh is not concerned with performance Jun 25 20:07:44 CFB is used by OpenVPN Jun 25 20:08:14 Rene_ Then it can do 150 Mbps with Shadowsocks I guess. Jun 25 20:08:32 I'm not sure how fast KCPtun is on it though. Jun 25 20:09:12 KCPtun speeds up the connection. Jun 25 20:09:12 heh, I wouldn't even bother these days Jun 25 20:09:26 vxlan and macsec due to low overhead when done in-kernel Jun 25 20:09:48 less so on non-x86 ofc Jun 25 20:09:54 wireguard runs in kernel too, and this is why it's not ported to Windows. Jun 25 20:09:59 and Mac I think. Jun 25 20:10:12 it can do yeah Jun 25 20:10:38 its not going to be as optimised for performance as crypto methods designed for zero latency though Jun 25 20:10:48 (and high throughput, ofc) Jun 25 20:11:31 kinda sucks that the embedded hardware crypto is so limited Jun 25 20:11:50 koops: wireguard has a Go userspace variant for Mac. Jun 25 20:11:52 If I want gigabit speeds over VPN do I need x86 CPU? Jun 25 20:11:55 cool. Jun 25 20:12:06 with x86 you can do 10s of gbps Jun 25 20:12:07 :P Jun 25 20:12:07 Already ported for Windows or not? Jun 25 20:12:22 nope Jun 25 20:12:29 I guess you could also do similar with pcie arm or mips Jun 25 20:12:36 koops: https://www.wireguard.com/install/ Jun 25 20:13:14 If I connect two computers directly, without any devices between them, do I need a regular Ethernet cable or a reverse one? Jun 25 20:13:31 depends on the hardware Jun 25 20:13:35 generally, no Jun 25 20:13:41 I want to make a 10Gbit/s link between my desktop and NAS. Jun 25 20:13:49 most sensible kit has auto mdi-x Jun 25 20:15:20 If there is no DHCP, I need to set IPs manually, right? Jun 25 20:15:39 if you want Jun 25 20:15:44 if link local isn't enough Jun 25 20:15:52 link local? Jun 25 20:16:00 well, "link local" Jun 25 20:17:12 Is it Linux-specific? Jun 25 20:17:15 no Jun 25 20:18:13 So if I simply connect two computers with Ethernet cable, they will be able to talk without configuration? Jun 25 20:19:28 depending on your choice of distro/os yeah, windows will sort itself out, linux will probably require dhcpcd or similar to generate linklocal Jun 25 20:19:39 so probably easier to just configure addresses Jun 25 20:19:43 or use v6 I guess Jun 25 20:21:00 So, if I will use IPv6 it will generate addresses automatically on any OS? Jun 25 20:21:34 by default, yes Jun 25 20:27:41 jwh: uhhh aes gcm is the same as aes ctr Jun 25 20:28:33 mangix: kinda Jun 25 20:28:59 my point it, aes ctr is the standard. the oddballs use CBC Jun 25 20:29:16 yeah Jun 25 20:29:30 CBC isn't really useful for network comms Jun 25 20:29:42 but its widely supported by crypto hardware heh Jun 25 20:30:15 but I guess by the same token, 3DES is even more widely supported Jun 25 20:30:16 :D Jun 25 20:30:20 hardware accelerates AES. CBC has nothing to do with it Jun 25 20:30:48 try telling that to vendor code Jun 25 20:32:00 not really that useful if you still need to do computations in software because it can't offload all stages Jun 25 20:33:38 you also see differing levels of acceleration depending on flavour Jun 25 20:34:09 hey! I'm using Reboot (17.01.4, r3560-79f57e422d) and I cannot use `telnet` command, why? busybox is installled and theoretically it should work Jun 25 20:38:51 at least re; cbc its gone in 1.3 Jun 25 20:47:42 guifipedro[m]: CONFIG_BUSYBOX_CONFIG_TELNET is off by default these days Jun 25 20:49:41 oh dear Jun 25 20:55:54 jow ping Jun 25 21:57:17 guifipedro[m]: (super duper late reply) y u no use ssh? Jun 25 22:00:35 drmr: some people use telnet for diagnostic testing of http / smtp.... Jun 25 22:01:19 oh, as a _client_. Jun 25 22:01:25 okay. Jun 25 22:02:40 but yes, it's not super necessary for most people :) Jun 25 22:03:01 interestingly, there's not even a package for telnet. Jun 25 22:03:16 you could always use nc, I guess. Jun 25 22:06:27 drmr: it's an optional chunk of busybox, as russel-- commented above Jun 25 22:07:01 karlp: yeah. I meant for those inclined to not compile their own images. Jun 25 22:15:28 drmr: I'm doing MTU testing. In one node I cannot run wget with https (openssl missing) and in the other I cannot do telnet (to open connection with remote point). I tried with netcat without success, I don't know how to proceed Jun 25 22:21:08 guifipedro[m]: problems with wget I get (I usually do curl), but what's the problem with netcat? I mean, it's not like it does anything much differently than a telnet client. Jun 25 22:23:02 telnet remains connection open, with `netcat gitlab 443` closes connection after hitting enter Jun 25 22:26:19 well, doing more tests and yes, it does very similar input/output Jun 25 22:26:34 uhm. it doensn Jun 25 22:26:36 I'm just tired today :) Jun 25 22:26:40 't immediately close the connection for me. Jun 25 22:29:16 but this test: `netcat localhost 22` is very similar to `telnet localhost 22` Jun 25 22:30:43 and/or with `telnet gmail-smtp-in.l.google.com. 25` and `netcat gmail-smtp-in.l.google.com. 25` Jun 25 22:36:00 Checking port 25 or 80 makes sense when you're using a text client. Use openssl s_client to test against SSL-enabled servers. Jun 25 22:40:11 If the square peg isn't being forced hard enough into the round hole to sheer off the corners, stuff a static binary in /jffs Jun 25 22:43:59 if you're going that far, just enable telnet again :) Jun 25 22:45:15 For MTU testing, try ping -M do -s 1472 victim.host to verify a 1500MTU Jun 25 22:46:09 longus_catus: yes, that is something you can do in debian, but this ping module is available in openwrt? Jun 25 22:46:45 to enable telnet again means to compile firmware,no ? :-) Jun 25 22:46:52 It's not available in the one in my image; victim should be whatever device you're questioning, since there are multiple problems assessing it from the inside-out Jun 25 22:47:00 and to put a static binary... depends on the binary if requires external dependencies Jun 25 22:48:31 I've found that static binaries generally have all the symbols referenced fully satisfied, unless they use dlopen() Jun 25 22:49:29 That might be exactly what you're saying, but it's hard to tell Jun 25 22:50:14 But, in my image, it has the 'openssl' binary, so if you want to try against a https target, use openssl s_client -connect host:port Jun 25 22:51:12 But, like, I'm assuming you've run ifconfig and have verified the device MTU, so this must be a more complex VPN issue Jun 25 23:13:20 thanks for the openssl suggestion (opkg install openssl-util, worked; on the other devices repositories are too old). No, it's a faulty or misconfigured device https://gitlab.com/guifi-exo/temba/issues/37 Jun 25 23:30:38 theres also socat :D Jun 25 23:32:20 but if you have curl installed anyway, you can use that (maybe, dunno if compile options preclude it) Jun 25 23:33:27 [jwh@pyxis libubox-git]$ curl telnet://openwrt.org:80 Jun 25 23:33:27 garbage input Jun 25 23:33:27 HTTP/1.1 400 Bad Request Jun 25 23:33:33 but any port, ofc Jun 25 23:34:54 root@fw1:~# curl telnet://openwrt.org:80 Jun 25 23:34:54 curl: (1) Protocol "telnet" not supported or disabled in libcurl Jun 25 23:34:56 aw Jun 26 00:25:59 build #49 of at91/sama5d4 is complete: Failure [failed sourceupload] Build details are at http://phase1.builds.lede-project.org/builders/at91%2Fsama5d4/builds/49 blamelist: Stijn Tintel , Yousong Zhou , Mathias Kresin , Franz Flasch Jun 26 01:09:51 build #528 of lantiq/ase is complete: Failure [failed sourceupload] Build details are at http://phase1.builds.lede-project.org/builders/lantiq%2Fase/builds/528 blamelist: Stijn Tintel , Yousong Zhou , Mathias Kresin , Franz Flasch Jun 26 01:12:04 build #48 of at91/sama5d2 is complete: Failure [failed sourceupload] Build details are at http://phase1.builds.lede-project.org/builders/at91%2Fsama5d2/builds/48 blamelist: Stijn Tintel , Yousong Zhou , Mathias Kresin , Franz Flasch **** ENDING LOGGING AT Tue Jun 26 03:00:02 2018