**** BEGIN LOGGING AT Wed Nov 06 03:00:06 2019 Nov 06 03:34:18 build #186 of arc770/generic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/arc770%2Fgeneric/builds/186 Nov 06 07:23:58 aparcar[m]: no Nov 06 07:24:19 aparcar[m]: I mena I have access to the mailman admin ui if you mean that, but I do not manage the infra Nov 06 07:27:01 jow: I'm just eager to check out lists.sr.ht as a lightweight patchwork replacement Nov 06 07:27:16 So the list could be added as a subscriber for temporary testing Nov 06 08:04:03 aparcar[m]: can't you just create some scratch email for that purpose and subscribe it to the list? Nov 06 08:05:52 ynezz as it's a list the srht list should be block to jot trigger an infinite back and forth, doesn't it? Nov 06 08:06:00 Maybe I have the wrong idea Nov 06 08:08:02 I would simply create some scratch gmail account, subscribed it to the openwrt-devel list, added forwarding rule to the lists.sr.ht Nov 06 08:11:16 Oh yea that sounds better Nov 06 08:15:44 BTW I'm wondering whats the difference between lists.sr.ht and patchwork Nov 06 08:16:48 I think it will have a tighter integration with git Nov 06 08:21:49 so you could actually git checkout branch-with-patchset ? Nov 06 08:22:00 I just can't find that info anywhere Nov 06 08:27:12 `git send-email --to ~ynezz/meh@lists.sr.ht some.patch` -> To: /home/ynezz/meh@lists.sr.ht (what could go wrong, right? :D) Nov 06 08:35:17 anyway, I can't see the difference between lists.sr.ht and patchwork, one would still probably need to poll the API (or use webhooks, which might be fragile) in order to create the git branch with the patch(set) in order to run the CI tasks Nov 06 08:46:27 woohoo, rc1. Great work :) Nov 06 08:52:25 weee Nov 06 08:52:54 blocktrron: Just noticed you ath9k patch for ar953x Nov 06 08:53:48 blocktrron: what is the behaviour you see before that one? Nov 06 08:54:13 blocktrron: purely, fyi, since the bump to mac80211 from 4.19 I'm seeing this often Nov 06 08:54:17 [751261.780158] ath: phy0: Timeout while waiting for nf to load: AR_PHY_AGC_CONTROL=0xd0d1a Nov 06 08:54:19 [751467.077256] ath: phy0: Timeout while waiting for nf to load: AR_PHY_AGC_CONTROL=0xd0d1a Nov 06 08:54:28 wondering if it's related Nov 06 08:55:16 xback: see this thread http://lists.infradead.org/pipermail/openwrt-devel/2019-November/019839.html Nov 06 08:56:29 ynezz: thanks :-) fast as usual Nov 06 09:03:01 preliminary changelog at https://openwrt.org/releases/19.07/changelog-19.07.0 Nov 06 09:03:08 someone needs to destill release notes out of that Nov 06 09:03:10 jow: thanks for tagging Nov 06 09:03:16 and everybody thanks for the hard work =) Nov 06 09:52:44 ynezz: thanks for being faster than my sleep :P Nov 06 09:53:07 god I hate pgp Nov 06 09:53:15 xback: seeing rc1 was branched, i wonder whether or not we should pick this patch for 19.07 Nov 06 09:53:26 apparently I am too stupid to import a secret key and use it Nov 06 09:53:38 It would keep the behavior in line with ar71xx, just seems correct Nov 06 09:54:26 I'll wait till evening and go for it then. Nov 06 11:29:59 The site is down : HY000 14 Can't change size of file (Errcode: 28 "No space left on device"):https://openwrt.org/docs/guide-user/installation/installation_methods/see_devicepage Nov 06 11:32:00 I am browsing it as we speak? Nov 06 11:36:01 it is up again Nov 06 11:36:38 is https://wikidevi.com/ working for you ? Nov 06 11:36:48 rPawel: wikidevi was shut down for good Nov 06 11:36:55 it does not exist anymore Nov 06 11:37:01 ah, thanks Nov 06 11:37:17 https://forum.openwrt.org/t/wikidevi-going-down-soon/45299 Nov 06 11:38:19 I am new to this community, I am about to convert my first router. Nov 06 11:39:04 unless I still decide to go the pfsense route :P Nov 06 11:39:53 (sidenote: #openwrt-devel is primarily geared towards developers, #openwrt is the general-purpose channel) Nov 06 11:40:33 I just noticed that we have one of these cookie consent things on openwrt.org Nov 06 11:41:28 understood, I will take my moans there :) and I will leave the mighty developers to themseves Nov 06 11:41:34 jow: a rather ugly one, but yes. is that a bad thing? Nov 06 11:42:06 takimata: uhm no, fine by me Nov 06 11:42:29 just never noted it, maybe because I am too conditioned to just click "ok, whatever" Nov 06 11:42:50 or because you are logged in. I didn't get it either until I used another browser to check. Nov 06 11:43:24 (or maybe it stores consent for a really long time.) Nov 06 11:44:05 rPawel: that wasn't implied, but if you have troubles installing or something end user related you will certainly reach more people there. Nov 06 11:45:01 lol, I am in a cheeky mood today, ps I am a dev too: scala/java :) Nov 06 11:46:54 Have a great day lads & lasses ! Nov 06 11:56:11 fyi, wiki is currently down for maintenance Nov 06 11:57:52 and it's back Nov 06 11:59:45 7mins, not bad :) Nov 06 12:07:16 quicker than i could e-mail to complain stintel :P Nov 06 12:08:10 well it was all DO .. I just shut it down, clicked in some gui, and turned it back on when it was finished Nov 06 12:08:21 cloud magic (tm) Nov 06 12:11:04 hehe Nov 06 12:59:39 morning Nov 06 13:52:31 moring, updating 19.07-rc now... Nov 06 13:53:51 * rr123 wishes irc has a vi mode Nov 06 14:14:10 https://bugs.openwrt.org/index.php?do=details&task_id=1800&pagenum=7 still is the case for me Nov 06 14:19:34 Hi people can some one ping me whan the RC is read for testing? I want to post it to twitter. Nov 06 14:19:41 ready* Nov 06 14:23:39 Tapper: as soon as the images are built i reckon ;) Nov 06 14:37:21 Borromini ok mate thanks just ping me pleas Nov 06 14:46:56 https_dns_proxy failed to compile: error: unknown argument: '-fhonour-copts' [clang-diagnostic-error]; https_dns_proxy-2018-04-23/src/dns_poller.c:4:10: error: 'ares.h' file not found [clang-diagnostic-error] Nov 06 14:47:02 19.07-rc1 Nov 06 14:47:16 blocktrron: https://git.openwrt.org/?p=openwrt/staging/blocktrron.git;a=commit;h=82ec3157e45a24ad21024aa5e43703e884231a4b Nov 06 14:47:35 blocktrron: warning, your SoB is present twice :-) Nov 06 14:55:29 rr123: is that an existing package bug? or just one to file on packages? Nov 06 14:59:04 not sure yet, submitted a bug to the package anyways Nov 06 15:14:41 xback: thanks, i've had a stupid alias for cherry-pick on this machine m( Nov 06 15:14:49 made this mistake more than one time already Nov 06 15:15:47 indeed stupid Git, can't figure out -s properly Nov 06 15:28:54 it usually happens to me while `git commit -vs --amend` the cherry-picked commit Nov 06 16:27:05 first time working with procd - i kinda like it Nov 06 16:30:41 fyi, new kernel bumps pushed to staging Nov 06 16:40:09 tmp/.config-package.in:104798:symbol PACKAGE_nfs-kernel-server depends on NFS_KERNEL_SERVER_V4 ; tmp/.config-package.in:104798:error: recursive dependency detected! Nov 06 16:40:28 not using this config but saw this warning about nfs-kernel-server on 19.07 Nov 06 16:42:00 BTW: are there any plans for the targets still at kernel 4.9 (ar7, orion, ixp4xx)? Just quickly looked at the history there, but it's only consisting of treewide changes until back to 2017. Nov 06 16:44:40 jow: I just had a look at http://openwrt.github.io/luci/jsapi/LuCI.rpc.html and it's really impressive and helpful Nov 06 16:44:49 thanks for taking the time to write so much documentation in one go! Nov 06 16:49:56 adrianschmutzler: IIRC all targets not converted to 4.19 (means lack of interest) are going to be removed after the next release is branched Nov 06 16:50:49 okay, and then also 18.06 with kernel 4.9 will die most probably Nov 06 16:51:38 hmm, I've just seen 4.14 patches from hauke for two of the targets in patchwork (from May ...) Nov 06 16:55:56 Good afternoon -- Two questions1) Any suggestions on how to fix `make menuconfig` showing ^@ in the terminal window and not responding to the [taab] key?2) What is the config option for including .config in the kernel? Nov 06 16:56:15 https://openwrt.github.io is 404 Nov 06 17:00:13 On (2) -- looking for /proc/config.gz Nov 06 17:04:31 rr123: that's how github pages worrk... you're not meant touse the root Nov 06 17:06:26 ok thanks Nov 06 17:07:50 Found it -- CONFIG_PACKAGE_kmod-ikconfig=y Nov 06 17:08:10 19.07 seems already using jow's js-cbi, without minify the js files and the rest of lua are larger in size Nov 06 17:08:54 without minify the js files, file sizes are larger than previous lua-mvc approach Nov 06 17:11:28 by comparing: du -s /www /usr/lib/lua; 19.07 is 2MB, 18.06 is 1MB, roughly. need minify and compare apple-to-apple Nov 06 17:14:10 Request URL: http://192.168.10.247/luci-static/resources/cbi.js?v=git-19.310.44720-c2be304 what is the usage of v here... Nov 06 18:20:17 jeffsf: for the @ thing with menuconfig, try "make -C scripts/config/ clean" followed by "make menuconfig" Nov 06 18:34:04 :+1: jow Nov 06 20:25:16 jow: I think it is cleaner to cherry-pick the weblate patches and apply them manually than merging the PRs Nov 06 21:14:20 Package dropbearconvert is missing dependencies for the following libraries: libz.so.1 Nov 06 21:16:00 but I do have CONFIG_DROPBEAR_ZLIB=y in .config, dropbear's makefile looks good, but somehow this failed for my clean 19.07 build. Nov 06 21:30:42 disabled CONFIG_DROPBEAR_ZLIB and can build 19.07 for ar71xx Nov 06 21:41:09 tp-link ac1750 C7 remains to be a popular model, amazon is selling a slightly cut-down version called A7, anyone tried c7 image on a7? there is no A7 options under menuconfig yet Nov 06 22:18:52 I would not recommend to buy the archer a7/ c7 new, if you already have them they'll remain solid options for many more years to come, but you can buy better devices for the same amount of money today Nov 06 22:19:40 that said, the c7 and the a7 are basically the same hardware, but require different firmware images (different h/w IDs, different partitioning and firmware formt), tplink_archer-a7-v5 is supported in master Nov 06 22:21:17 flashing a c7 image on the a7 would a) hopefully not work, b) break the device Nov 06 22:25:01 what's the "better option for the money" right now? Nov 06 22:25:35 karlp: imho ipq40xx (e.g. Linksys EA6350v3, ZyXEL NBG6617, AVM 4040, ...) Nov 06 22:31:45 karlp: these days multi-core imho does make sense (even if one core is clogged with work (ppp, VPN, etc.), the device still remains snappy) - and qca4018/ qca4019 wlan (yes, only two streams, but wave2) is better than QCA9880 with three streams) Nov 06 22:40:29 the EA6350v3 is slightly more expensive than the Archer C7 Nov 06 22:40:34 but I agree that the hardware is better Nov 06 22:41:51 btw, I'm also impressed with the performance of ramips/mt7620, it can almost forward a gigabit without flow offloading Nov 06 22:41:53 zorun: most ipq40xx have gotten more expensive over the last 3-4 months (no idea why), but until ~summer, the archer c7 sold for ~70 EUR new, the nbg6617 and avm 4040 for ~65 EUR. and the ea6360v3 was on sale in the UK for 35 GBP Nov 06 22:42:22 imho the archer c7 is just a bit overprised these days, both new and used Nov 06 22:42:38 oh ok, I bought my EA6350 for 78 € Nov 06 22:43:07 the big problem with the ea6350, apparently there now also is a v4 variant (mt7621 based) Nov 06 22:43:08 and the C7 is 70 € indeed Nov 06 22:43:15 ah Nov 06 22:43:18 v4 then? Nov 06 22:43:26 not supported yet Nov 06 22:43:39 (and personally I'd favour ipq40xx over mt7621) Nov 06 22:44:02 finding a reliable and long-term router model is always a pain... Nov 06 22:44:25 the C7 has been nice in that regards (except the high price) Nov 06 22:44:38 pkgadd: because of the number of CPU cores? Nov 06 22:45:15 I have an mt7621, and it can do my gige wired nicely, but it's 2.4gig and 5gig performance is not crash hot. Nov 06 22:45:26 zorun: everything, actually. 4 cores vs 2 cores (yes, in terms of routing throughput mt7621 has the upper hand), but also because of the current state of their wlan driver Nov 06 22:46:40 zorun: I'd like to add mt7621 to my 'collection', but I'd only be willing to pay (significantly) less for it than I would for ipq40xx Nov 06 22:48:22 given that mips seems to be dying, I'd also prefer ARMv7 (ipq40xx) over mt7621 (mips) on that ground alone Nov 06 22:49:06 ok Nov 06 22:49:39 if you take into account power usage, I think mt7621 has the upper hand Nov 06 22:49:52 the ipq40xx device I tried got really hot Nov 06 22:50:14 perhaps, yes - not sure if that's much of a difference though Nov 06 22:53:27 mt7622 would be more interesting, but options are very limited there Nov 06 23:03:03 pkgadd: didn't the EA8350 have a laughable low limit on the kernel size? Nov 06 23:04:02 https://github.com/openwrt/openwrt/pull/1716#discussion_r245342344 Nov 06 23:05:31 blocktrron: 3 MB, as it seems kernel@0 rootfs@300000 Nov 06 23:05:41 urgh Nov 06 23:06:24 at least they've shared the same love with zyxel while rolling the dive for their partition layouts Nov 06 23:06:46 kernel@0 { reg = <0x00000000 0x02800000>; }; rootfs@300000 { reg = <0x00300000 0x02500000>; }; so the usual Linksys typicaly overlapping partitions, should be possible to extend that Nov 06 23:07:12 my zyxel nbg6817 has at least 4 MB (GPT on eMMC) Nov 06 23:07:34 yeah, if you build your custom image as this would include altering the partition layout Nov 06 23:07:45 nasty for sure Nov 06 23:07:54 they did the same stuff to the nbg6617 (kernel after rootfs with 4M Kernel partition) Nov 06 23:08:16 So the AVM 4040 is still somewhat your best bet for ipq40xx Nov 06 23:08:32 blocktrron: that is not an issue on the nbg6817 (yet), https://openwrt.org/toh/zyxel/nbg6817#flash_layout Nov 06 23:09:00 the version # kills me, same model number with different version could mean totally different CPU/RAM/Flash, can vendors do meaningful semver release/labelling Nov 06 23:09:33 7621 is great but its popluar 2.4G companion chip 7306e sucks, mt76 could not fix that either Nov 06 23:09:35 pkgadd: i was talking about the nbg6617, not the nbg6817 Nov 06 23:09:52 i have a collection of zbt1326 here collecting dusts, due to crappy 7306e chip on it Nov 06 23:10:14 However, i'm still somewhat puzzled why we limit ourselves to 4M kernel size, as extending it means breaking sysupgrades Nov 06 23:11:01 At least for NAND devices, i would go for 8M to buy myself a large-enough amount of time Nov 06 23:11:25 yes, it's sad that vendors do limit themselves that way Nov 06 23:11:43 Or have the kernel in a ubi volume, but apart from Riverbed, AVM and ASUS no one seems how to do that Nov 06 23:13:45 for practical intents and purposes, AVM 4040, (7520), 7530 would be the most interesting ipq40xx devices for me; nbg6617 has gotten rare/ expensive. the ea6350 was just interesting at its price point (now with the mt7621 based v4, that's no longer interesting to me) Nov 06 23:14:27 i assume it's not that nice to get AVM hardware outside of germany Nov 06 23:14:57 interesting enough, they're selling quite well in australia, but aside from that mostly germany and nearby EU Nov 06 23:15:31 I like how they've just slammed 128M of NANd flash into their latest repeater and sell it for 50 bucks Nov 06 23:15:42 a chip designer from ericsson told me today that RISC V is the future...said ARM charged a leg and an arm and it adds up for large volume production, for multi-core chip, each core needs a separate license fee Nov 06 23:16:10 basically arm can charge at will as there is no alternatives, thus big guns are betting on RISC V heavily already Nov 06 23:17:03 Maybe we can get some nice ipq40xx / ipq806x hardware from Aruba's new lineup for cheap, but that's again only access points Nov 06 23:17:06 rr123: wake me up when riscv is actually there (keep in mind, fees for non-core-CPU IP blocks, RAM controller, storage, networking, gpu, etc. remain), in interesting devices Nov 06 23:18:37 on othe other hand samsung just dismissed its whole cpu department and will use arm fully for the future, it's not easy to do your core Nov 06 23:20:01 will check AC7150 A7 from master and hopefully backport to 1907 as I just bought a cheap piece of that device Nov 06 23:20:21 I'm keeping my eyes open for 'cheap' used ipq40xx/ ipq806x devices (nbg6817, (nbg6816), (r7500), r7500v2, r7800, (archer c2600), (ea8500), (tew-827dru v1), nbg6617, avm 4040/ (7520)/ 7530, ea6350, ea8300) myself, for experimenting (so cheap is king), but I haven't really found anything within my allocated budget 'limits' so far Nov 06 23:20:38 for me, AC1750 is ready fast more that enough for the whole house, i mean, who needs wave2, what's for? Nov 06 23:20:59 it's not like i'm going to watch 8K video on my tiny cellphone screen via wifi anytime Nov 06 23:21:12 ex7500, ex7700 or ex8000 could also be interesting Nov 06 23:23:28 chatted with possible 5G application, are we going do the dizzy VR gaming on the go? I never thought 4G is not enough for 99% of the population. Nov 06 23:23:48 rr123: in my practical tests, ipq4019 wlan is faster as qca9880 (bthub5) and provides slightly better throughput over the distance Nov 06 23:24:30 and beyond 100 MBit/s WAN-to-LAN throughput, ipq40xx wins hands down Nov 06 23:24:41 same for VPN Nov 06 23:24:58 how is vpn throughput? Nov 06 23:25:06 having VPN 'confined' to one core is also better Nov 06 23:25:39 rr123: https://forum.openwrt.org/t/comparative-throughput-testing-including-nat-sqm-wireguard-and-openvpn/44724 Nov 06 23:25:42 you mean 'taskset' it? Nov 06 23:26:53 rr123: no, but VPN is usually single threaded, so on a fast link, under pressure, it can still only make use of a single core (not always the same, but 1 core at most), leaving the other cores for the other important tasks (IRQs, PPP, WLAN, ...) Nov 06 23:27:31 pkgadd: depends on cpu, some with crypto engine will offload vpn Nov 06 23:27:56 never tried irq/process affinity on a router yet, as bandwidth is not a priority yet here Nov 06 23:27:57 rr123: sure, but that's a different story (and needs special configuration) Nov 06 23:28:07 basically the bandwidth is good enough for me so far Nov 06 23:30:18 rr123: with PPPoE (and no SQM) ath79 is good for ~150 MBit/s, it's easy to hit those limits these days (I'm on 100/40 MBit/s now, but wuill move to 400/200 within the next 6-12 months) Nov 06 23:31:28 anyone ever found a usb ttl where the USB part can be easily mounted to a panel? Trying to find a nice way to bring a serial port to the outside of a case. Nov 06 23:32:50 pkgadd: talking about ath79, 19.07 still does ar71xx correct, my build under bin/target still uses ar71xx at least Nov 06 23:33:44 greearb_: I've had my eyes on https://www.ebay.de/itm/x/152679798450 but I haven't actually bought any of those yet (you can get variants with the pins soldered in and variants without) Nov 06 23:34:12 rr123: yes, but I've been on ath79 since the beginning of the year Nov 06 23:35:38 The WPQ872 doesn't come with a case, so I am working on building one. It has UART pins on it already, and on the back of the board, so I'll need a cable to run from that to the panel. I'm thinking normal usb TTL plus this: https://www.amazon.com/gp/product/B01N6ERRX7/ref=ox_sc_act_title_1?smid=A19EHJUVB4I00W&psc=1 Nov 06 23:35:48 but would be cleaner if I didn't need the bulkhead adapter Nov 06 23:37:36 I know a few people who just use 3.5mm (or 2.5mm, whatever floats your boat) TRS connectors... Nov 06 23:39:29 pkgadd: how ath79 is much better, searched for some doc to try out, found a few forum posts only, AC1750 seems supported Nov 06 23:40:06 rr123: ar71xx will cease to exist after 19.07, it's already gone from the master buildbots Nov 06 23:41:26 any quick-start thing to use ath79, or i have to read all those forum posts to figure out... Nov 06 23:42:13 linux/generic/config-4.14:# CONFIG_ATH79 is not set Nov 06 23:42:38 might be just kernel_menuconfig to turn ATH79 option? Nov 06 23:43:06 greearb_: just in general (STA uses, maybe dabbling into monitor mode a little, obviously no AP mode), how do you get along with intel ax200 in practice - is it worth buying (remembering your thread on linux-wireless)? (I'm looking for a 802.11ax card for my desktop, so m.2 <--> desktop PCIe adaptor) Nov 06 23:43:53 it crashes in upload direction, but driver usually recovers. In general, seems nice enough...we ran 8 in a chassis in overnight test and system survived. Nov 06 23:44:16 rr123: if your device is already supported on ath79 (the archer c7/ a7 are), you can just switch the target arch and build/ flash (not keeping configuration is necessary, -F | --force might be needed in some cases) Nov 06 23:44:25 thanks Nov 06 23:47:00 will try, just checked rbm33g is supported in 19.07 now, nice Nov 06 23:48:06 yes saw ath79 on cpu menu **** ENDING LOGGING AT Thu Nov 07 02:59:57 2019