**** BEGIN LOGGING AT Sun Jun 24 03:00:00 2018 Jun 24 03:51:59 https://downloads.openwrt.org/snapshots/targets/ This item does not seem to be updated synchronously Jun 24 08:03:52 movi: unfortunately not Jun 24 08:07:01 xdarklight: its not migrated and there is no easy way to do so Jun 24 08:09:40 Hi blogic hi Jun 24 08:10:02 Quick q about 186 Jun 24 08:10:14 shot Jun 24 08:11:04 18.6 you know the 4 commits for backports to mac80211? Jun 24 08:11:23 are they ment to go in to master to? Jun 24 08:11:24 ath10k fw, powersave, leds and summin else Jun 24 08:11:28 those ? Jun 24 08:11:33 they are in master Jun 24 08:11:58 the stuff from mac80211: backport "brcmfmac: cleanup and some rework" from 4.17 Jun 24 08:12:17 hash ? Jun 24 08:12:30 0f1a1489a6c584d7de6680a65b457acf328818bd ? Jun 24 08:12:33 https://github.com/openwrt/openwrt/commit/8cfd43d1d65aa70dff06ef2f63186a4c4285bccc Jun 24 08:12:57 rmilecki: ^^ Jun 24 08:13:01 and the next three Jun 24 08:13:07 hey Jun 24 08:13:08 rmilecki: that should be in trunk aswell really Jun 24 08:13:24 did you maybe simply forget to push ? Jun 24 08:13:44 uh Jun 24 08:13:49 let me check... i don't think so Jun 24 08:14:21 Tapper: the commit even says "(cherry picked from commit 0da9303)" Jun 24 08:14:26 Tapper: check for that commit Jun 24 08:14:29 it's in the master branch Jun 24 08:14:38 rmilecki: correct, sorry too early i missed that Jun 24 08:14:45 blogic: no problem! Jun 24 08:14:45 Ok just wanted to make shure sorry Jun 24 08:36:03 blogic Is jow the best one to ask about a change to luci? I just would like some hading tags aded to help people with screenreaders. Jun 24 08:36:14 heading* Jun 24 08:37:24 some of the things on the Status page should be tables to. Jun 24 08:38:16 I know they are tables to a sited person, but they don't show up as tables to a screen reader. Jun 24 08:39:24 Should i make a new Issue? Jun 24 08:40:21 that's going to be quite an undertaking. lots of LuCI's output is hardcoded, unfortunately. Jun 24 08:40:49 No it's just in the html code just tags I think Jun 24 08:41:20 that's what I mean, one would have to go into each and every file and change the output HTML. Jun 24 08:41:22 the tables work in 15.1 I am shure Jun 24 09:13:55 Improvements for the usage of screen readers! #1907 Jun 24 09:13:59 https://github.com/openwrt/luci/issues/1907 Jun 24 09:19:33 Tapper: he is Jun 24 09:20:20 K thanks. I have rote it up now any so it does not get lost in the chatter. :D Jun 24 09:20:33 anyway* Jun 24 09:30:24 morning Jun 24 09:32:45 today is special day Jun 24 09:33:03 today is it 6 months to christmas Jun 24 09:33:24 time to eat the left over xmas cookies then Jun 24 09:34:29 yes, my wife is preparing meal we have for christmas. Cookies, christmas songs and chritsmas tree was not permited Jun 24 09:35:31 ultito OMG pleas don't lol I have 5 kids so my bank account tacks a rite kicking! Jun 24 09:36:16 Tapper: I have only one kid, but I know what you mean Jun 24 09:37:45 we have on June 24th (or as close as possible) for few years just a small reminder for christmas by a meal we have on christmas eve :-) Jun 24 09:38:32 This year I did not take down christmas lights on my house....hmmm nice idea :-) Jun 24 09:44:22 Hey, since this morning my device (Netgear WNDR4300) won't accept ramdisk images saying "Uncompressing Kernel Image ... ERROR: LzmaDecode.c, 547". This started after I included LuCI in the image, but after removing luci the error stays. Jun 24 09:47:40 https://www.redbubble.com/people/softwareguru/works/16328367-ich-bin-ein-coder-was-ist-deine-supermacht?cat_context=u-prints&grid_pos=11&p=metal-print&rbs=29bf07ae-90ef-41ba-91f9-abd97e4c0822&ref=shop_grid Jun 24 09:47:44 lulz Jun 24 09:52:04 mangix: that's 8 degrees higher than inactive mwlwifi card :) Jun 24 10:24:16 Hertog_Jan: use a different tftp addr Jun 24 10:24:34 Hertog_Jan: you load the initramfs to location X and then bootm unpacks it to Y Jun 24 10:24:45 it will use Y+Zmb Jun 24 10:24:54 and X is most likely inside Y+Z range Jun 24 10:25:02 Oh right that makes sense Jun 24 10:25:07 what tftpboot addr are you using ? Jun 24 10:25:09 man ath10k is junk Jun 24 10:25:20 try 0x82000000 that is 32 mb into the window Jun 24 10:25:25 stintel: no shit ? Jun 24 10:25:33 the C in QCA stands for crack Jun 24 10:25:34 tftpboot 80800000 openwrt-ath79-nand-netgear_wndr4300-initramfs-kernel.bin Jun 24 10:25:40 blogic: ;) Jun 24 10:25:54 Hertog_Jan: yep, that is 8 MB into the window Jun 24 10:25:59 use 0x82000000 Jun 24 10:26:14 stintel: why particularly do you say its shitty today ? Jun 24 10:26:20 OK, thanks a bunch Jun 24 10:26:34 ath10k is an example that shittiness is realtiv Jun 24 10:26:49 the weather is shitty ... but its better than ath10k Jun 24 10:27:01 thus making shittiness relativ ... Q.E.D. Jun 24 10:28:30 blogic: am in Belgium for a couple of weeks, the AP (DAP-2695-A1) is up for 1 day, and athk10k goes like this: https://pastebin.com/8JEqVhj3 Jun 24 10:28:54 that is pretty shitty indeed Jun 24 10:29:08 worst part now is that it stays broadcasting ssid and clients try to connect to it and fail Jun 24 10:29:15 for some reason they didn't fall back to 2.4GHz either Jun 24 10:29:26 i have an idea ... Jun 24 10:29:30 disable the 5ghz radio Jun 24 10:29:31 hammertime ? Jun 24 10:29:33 looks like seriously buggy cod Jun 24 10:29:34 e Jun 24 10:30:13 blogic: starting to feel like taking my QCA based APs to the shooting range next time I go Jun 24 10:32:43 stintel: I see the same bug on a Home Hub 5A with QCA9880-BR4A - which wifi chipset does your device use? Jun 24 10:33:29 xdarklight: see https://bugs.lede-project.org/index.php?do=details&task_id=333 Jun 24 10:34:01 still the same bug, also https://bugzilla.kernel.org/show_bug.cgi?id=188201 Jun 24 10:34:08 nobody gives a f* Jun 24 10:34:38 so yours has a QCA9880 as well Jun 24 10:34:55 and my experiments with -ct firmware ... looked more stable, but bandwidth is useless Jun 24 10:35:05 couldn't even stream 1 video stream Jun 24 10:35:34 did you use -ct firmware + -ct driver or -ct firmware + default driver? Jun 24 10:35:40 all combinations Jun 24 10:36:10 I have the same problem with a BTHH5A Jun 24 10:36:46 we could try to reload the driver when "ailed to set beacon mode for vdev 0: -11" happens Jun 24 10:36:46 I built a -ct firmware + -ct driver image a few hours ago because I heard that there's some debug logging with that firmware + driver combination... maybe Ben can make any sense of the errors (in case any error shows up...) Jun 24 10:37:30 extend ath10k to be able to recover from this problem Jun 24 10:38:27 is it possible to trigegr this, for me it normally happens after 1 to 5 days Jun 24 10:39:04 well I have received my Gemini PDA Jun 24 10:39:10 Mediatek Helios X27 Jun 24 10:39:20 my phone is Huawei with Kirin 970 Jun 24 10:39:30 and laptop is Intel 8265 Jun 24 10:39:35 2 very unusual devices Jun 24 10:39:41 the combo seems to trigger it easily Jun 24 10:40:57 oh and maybe my tablet (exynos) Jun 24 10:41:13 I mostly have two or three clients: another QCA9880 (as client), mobile phone with snapdragon 625 and another mobile phone (xperia z3 compact, no idea what wifi chip they use) Jun 24 10:41:30 that seems to trigger it as well Jun 24 10:41:58 I guess I should try -ct firmware again Jun 24 10:42:10 at least Ben seems to respond to problems Jun 24 10:42:15 QCA are dead silent Jun 24 10:42:45 they're probably too busy creating new chips with more shitty firmware Jun 24 10:44:26 I have PCs with Intel 6300 and 7260, Samsung S4 with BCM4335 , Samsung TV (I do notn know the wifi chip) Jun 24 10:47:00 stintel: That Gemini pad looks very nice ! how do you like the keyboard ? which OS are you runnin ? Jun 24 10:54:33 stintel: 'too busy... with more shitty firmware' You know that if they made a working product, they'd have trouble selling the next model. Jun 24 11:08:17 stintel: -ct driver and firmware resolved those issues for me Jun 24 11:09:13 stintel: HTT firmware worked poorly for me but may work better for you. Jun 24 11:11:38 Monkeh: do tell Jun 24 11:12:08 ..? Jun 24 11:21:41 hi Jun 24 11:44:59 Monkeh: is ath10k-ct more stable overall that vanilla? Jun 24 11:45:30 i have a Mac here which seems to positively hate it Jun 24 11:45:38 For me, yes. Jun 24 11:50:02 blogic: sorry to bother again, could you maybe help me find out where the wan port is defined in the ar71xx target for the netgear wndr4300? Jun 24 11:50:26 I've been looking through mach-wndr4300.c, but I don't think it is in there Jun 24 11:51:11 Or maybe it is setup in ath79_setup_ar934x_eth_cfg()? Jun 24 12:20:55 On a hunch I tried using the mdio device where the lan phy is connected, that seems to work :) Jun 24 16:03:40 Monkeh: which firmware do you use with it? HTT or non-HTT? Jun 24 16:05:09 htt Jun 24 16:07:38 thanks. I remember Ben being on IRC at some point, but I guess not anymore Jun 24 16:11:23 movi: he might just be on vacation or sth. he hung out here regularly Jun 24 17:11:17 nope, -ct doesn't fix my problem :/ Jun 24 17:49:22 mangix: Wistron NeWeb is actually a quite popular Taiwanese company with public shares. Jun 24 17:49:41 I was worried it's a no-name private Chinese company because of the name. Jun 24 17:51:06 For some reason a company located in Hong Kong or Taiwan doesn't make me as nervous. Jun 24 17:51:46 And the fact that the company has publicly traded shares and it's quite big and not a complete no-name. Jun 24 17:59:00 It's amusing that you fear the no-names not the big obvious target brands Jun 24 18:00:03 A company with no name has nothing to lose in case of discovery. Jun 24 18:00:22 It can simply close and reopen by another name. Jun 24 18:07:17 hello does anyone know of a place i can get an x86 virtual machine of current lede 17.01.4? Jun 24 18:08:08 xpressh: LEDE merged back into OpenWRT Jun 24 18:09:46 https://downloads.lede-project.org/releases/17.01.4/targets/x86/64/lede-17.01.4-x86-64-combined-ext4.img.gz Jun 24 18:09:55 This is a disk image. Jun 24 18:10:11 koops: awesome! ty! Jun 24 18:10:22 np Jun 24 18:10:29 didn't realize it was sitting right in the repo Jun 24 18:13:35 do you know if open-vm-tools will ever be added? Jun 24 18:14:30 I don't know. Jun 24 19:38:46 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Jun 24 20:30:13 How popular/trustworthy GL.iNet is? Jun 24 20:39:21 what is that? Jun 24 20:44:53 Router manufacturer. Jun 24 20:46:22 so my question is answer for you :-) Jun 24 20:46:37 devices look ok, trustworthiness probably doesnt matter if you are putting openwrt on it Jun 24 20:52:09 jwh: What about a U-Boot backdoor? Jun 24 20:52:14 Or a microcode one? Jun 24 20:52:30 Both were found in generic Aliexpress routers. Jun 24 20:53:08 why would they do that Jun 24 20:53:25 To get some additional profit most likely. Jun 24 20:53:48 I will find a quote on microcode ones in a sec. Jun 24 20:55:18 * ldir ipflibbydibbys Jun 24 21:02:50 glinet is well behaved, submits patches, provides support. Jun 24 21:03:10 this mythical backdoor won't have come from them, if you insist on believing it Jun 24 21:05:03 ZBT? Jun 24 21:05:32 I only see ZBT devices using MediaTek MT7621AT though. Jun 24 21:05:51 This SoC doesn't have proper multithreading. Jun 24 21:06:04 Running two things at once may freeze your router. Jun 24 21:06:49 karlp: I'm sure someone mentioned a CPU microcode backdoor in some generic device. Jun 24 21:07:14 yeah, in x86 northbridges, developed in the land of the free. Jun 24 21:07:57 no, not that. Jun 24 21:08:14 It was in an unbranded router from Aliexpress if I remember correctly. Jun 24 21:08:16 wasnt really a backdoor anyway, just questionable design Jun 24 21:08:48 no no, if you buy it from amazon, it's clean, but if you buy it from ali, it's definitely backdoored.... Jun 24 21:08:58 wrong. Jun 24 21:09:12 If you buy a generic device you have no idea what can be inside. Jun 24 21:09:13 what control does uboot really have once its loaded the application anyway Jun 24 21:09:29 well, it could tamper with the application while loading it. Jun 24 21:09:42 it isnt like x86 bios, right? Jun 24 21:09:48 isnt running all the time Jun 24 21:10:05 it's ACPI and ME that's running at all times. Jun 24 21:10:16 yes, which only arm64 has Jun 24 21:10:19 Not the BIOS itself. Jun 24 21:10:25 and they dont use uboot Jun 24 21:10:27 umm Jun 24 21:10:38 bios is running Jun 24 21:10:59 BIOS contains some Windows DLLs too for some reason. Jun 24 21:11:09 ok tgen Jun 24 21:11:11 then Jun 24 21:12:35 BIOS can contain windows stuff that's copied in at installation time. that's not a big secret (and they are not trying to hide it). Jun 24 21:13:31 BIOS can do all sorts of nasty stuff without these files actually. Jun 24 21:14:15 The most obvious thing is to put something dodgy in these files, but it's more clever to not do that, and instead tamper with the OS in memory. Jun 24 21:16:24 I think BIOS updates on modern mobos are signed, but the actual flash chip contents are not. Jun 24 21:16:52 so your OpenWrt-related question would be? Jun 24 21:17:40 So, the risks of getting a backdoored router are not less severe than a backdoored mobo. Jun 24 21:17:54 q.e.d. Jun 24 21:18:27 you'll be fine unless you buy your tinfoil hat from Aliexpress. Jun 24 21:18:34 * karlp hi5s drmr Jun 24 21:18:40 >Not the BIOS itself. Jun 24 21:18:40 I think OS can call BIOS using interrupts though. Jun 24 21:18:57 * ldir chuckles Jun 24 21:19:18 I guess a random seller from Amazon is equally bad. Jun 24 21:19:30 Aliexpress is just an example. Jun 24 21:20:25 i suggest you read some architecture docs Jun 24 21:20:36 Which ones exactly? Jun 24 21:20:45 all of tgen Jun 24 21:20:48 them Jun 24 21:21:41 * drmr personally likes the Gothic Revival. Jun 24 21:22:08 x86? Maybe ARM? Jun 24 21:22:19 x86 definitely Jun 24 21:26:33 I mostly develop in JS to be honest. I may be wrong about x86 internals. Jun 24 21:27:21 no shit Jun 24 21:29:19 How to understand this - packets sent through the WinMTR/MTR interval less than a second begins to miss answers. The router is not coping or? https://s33.postimg.cc/r6ci02gxb/Image_1.gif 0.2 msec Jun 24 21:29:20 drmr: bios *updates* can include dll's and similar in the archive, but the bios image itself cannot. Jun 24 21:29:33 lede pi 2 Jun 24 21:29:35 EqUaTe: of course it can. Jun 24 21:30:18 EqUaTe: where do the DLLs go then? Jun 24 21:30:37 qwefytuiityty: icmp tends to get lowest priority, so routers will drop when it gets too frequent. short answer is that it's dropping them in favour of actually doing normal stuff. if you monitor the load while you do that you'll find that the CPU is v high. Jun 24 21:30:42 drmr: no, it really can't. Jun 24 21:30:45 Are they installed to Windows instantly when you update? Jun 24 21:30:45 0.9 - nterval less than a second begins to miss answers Jun 24 21:30:50 koops: they're used as part of the flash tool. Jun 24 21:31:03 so they're usually loaded into RAM to flash the update, and that's it. Jun 24 21:31:22 EqUaTe: BIOS can contain drivers and other files windows will automatically install. Jun 24 21:31:26 the BIOS chips don't have the capacity to contain files Jun 24 21:31:34 drmr: no, it really can't Jun 24 21:31:35 EqUaTe: Lenovo famously used it to smuggle in their "Lenovo Service Engine" Jun 24 21:31:42 that wasn't in the bios.. Jun 24 21:31:42 this not only witch lede. Other router not lede/openwrt the same Jun 24 21:31:49 with Jun 24 21:32:13 I actually consider building a router out of a x86 mini-ITX mobo. Jun 24 21:32:17 qwefytuiityty: yes. you'll also see that behaviour on the type of router used at the core of your ISPs network. Jun 24 21:32:18 EqUaTe: do I _really_ need to look up the mechanism for you now? Jun 24 21:33:09 drmr: one, BIOS has no way to even access filesystems, two, you're getting confused with EFI (EFI != BIOS), 3, the files themselves are stored in a special partition on local storage that is often hidden from end users. Jun 24 21:33:13 drmr: i think you mean the mechanism to supply drivers at install time Jun 24 21:33:21 which has nothing to do with bios Jun 24 21:33:41 EqUaTe: it's the other way around. windows has a mechanism that looks up and installs files contained in the BIOS. I'm not making this up, really. Jun 24 21:33:54 no, it really doesn't Jun 24 21:34:03 hidden partitipns, not from the running bios image Jun 24 21:34:05 how silly Jun 24 21:34:10 partitions Jun 24 21:34:18 you clearly don't understand what a BIOS is/was. Jun 24 21:34:34 EqUaTe: okay, there it is. it is called "Windows Platform Binary Table" - http://download.microsoft.com/download/8/A/2/8A2FB72D-9B96-4E2D-A559-4A27CF905A80/windows-platform-binary-table.docx Jun 24 21:35:00 sorry, i'm not going to open a random word document, even if it is on ms.com Jun 24 21:35:02 I guess I will eventually grow out of worrying about backdoors so much. Jun 24 21:35:15 however, a BIOS ACPI table does *NOT* contain any files. Jun 24 21:35:21 it can contain *pointers* Jun 24 21:35:31 EqUaTe: But you are fine viewing random images over HTTP? Jun 24 21:35:55 pb is uefi only Jun 24 21:35:57 koops: it's a lot harder to sandbox a word processor than a browser. Jun 24 21:36:14 EqUaTe: uhm, I'll take it back, I conflated this with EFI. sorry. Jun 24 21:36:21 Any sandbox can be broken out of by malware. Jun 24 21:36:23 also - none of these is even slightly relevant to openwrt Jun 24 21:36:33 its also not drivers, has limited use Jun 24 21:36:33 that is true. Jun 24 21:36:48 Assume that if malware gets executed in any number of sandboxing layers, it's over. Jun 24 21:37:04 koops: hence not opening suspect files. Jun 24 21:37:21 in other words, applying basic levels of caution. Jun 24 21:37:25 There is no 100% way to sandbox any code, mostly because of CPU vulnerabilities. Jun 24 21:37:37 row hammer etc. Jun 24 21:37:41 NSS. Jun 24 21:37:46 lol Jun 24 21:37:49 okay, guys, I think we all got a bit carried away here. Jun 24 21:37:58 can we return on topic? Jun 24 21:38:10 as much fun as this is, yes Jun 24 21:38:13 heh Jun 24 21:38:29 * EqUaTe fades back into the background Jun 24 21:38:46 EqUaTe: (sorry again, my bad.) Jun 24 21:38:55 Even most advanced sandbox can't hide the underlying CPU's vulns. (last post) Jun 24 21:39:42 koops: go look up what NSS might stand for. i'll give you a hint, it's an acronym that references a detective about whom sir arthur conan doyle wrote. Jun 24 21:40:09 I got the reference. Jun 24 21:42:06 and I just noticed that I forgot everything I ever knew about rsyncd. Jun 24 21:43:02 This is what I'm concerned about: I buy a random device, plug it in, and within minutes the contained malware compromises all my devices' software and hardware. Then it implants itself in my web accounts and even if I swap all hardware it won't be removed because I will eventually have to log into the compromised account. Jun 24 21:43:18 I guess that happens if you set up a sync system that runs flawlessly for almost a decade. Jun 24 21:43:41 Isn't rsync very slow? Jun 24 21:44:07 koops: then maybe don't buy a random device. Jun 24 21:44:26 crisis averted, babies safe. Jun 24 21:44:42 and kittens. Jun 24 21:44:58 kittens are never safe. but they come with extra lives. Jun 24 21:45:39 In China kittens are never safe from becoming food. Jun 24 21:45:54 In Korea (both north and south) too. Jun 24 21:47:00 Linksys WRT3200ACM Jun 24 21:47:00 Linksys WRT1900AC Jun 24 21:47:00 Linksys WRT1900ACS Jun 24 21:47:00 Linksys WRT1900AC Jun 24 21:47:00 TP-Link Archer C7 AC1750 Jun 24 21:47:01 D-Link DIR-869 Jun 24 21:47:01 ZyXEL NBG6616 Jun 24 21:47:02 Asus RT-AC58U Jun 24 21:47:02 GL.iNet GL-B1300 Jun 24 21:47:12 I need help choosing between these. Jun 24 21:47:48 I got this list by excluding as many routers as possible. Jun 24 21:48:04 I guess any of these are a decent choice for OpenWRT. Jun 24 21:48:41 Maybe exclude GL.iNet for only having 2 LAN ports and ZyXEL NBG6616 for look. Jun 24 21:48:57 But, I looked and the load on the CPU cores is very small 0.1-1% when i used with interval wimtr 0.2 - 0.9 msec Jun 24 21:49:36 https://www.amazon.com/ZyXEL-Simultaneous-Dual-Band-Wireless-Support/dp/B00XQZNZ1C Jun 24 21:49:39 No longer sold too. Jun 24 21:50:19 htop Jun 24 21:58:59 http://www.amazon.com/dp/B01LXL1AR8 Jun 24 21:59:07 Is this OpenWRT-compatible? Jun 24 21:59:21 Seems identical to ASUS RT-AC58U. Jun 24 22:02:37 send 238 192.168.1.12, next ip send 1219 https://ufile.io/g6wdu Jun 24 22:03:07 interval 0.2 msec Jun 24 22:04:28 sent Jun 24 22:05:34 qwefytuiityty What are you doing? Jun 24 22:06:57 wimtr start trace to ip with option 0.2 msec interval Jun 24 22:07:16 winmtr or linux mtr Jun 24 22:09:22 0% if option standart 1 sec , 0.9 msec have loss % Jun 24 22:09:55 Is there upstream work on Marvell 88W8964? Jun 24 22:10:15 loss if interval 0.2 - 0.9 Jun 24 22:11:34 not loss packages, no answers Jun 24 22:12:35 you have lede and can see Jun 24 22:13:12 koops: I wouldn't buy ar71xx new anymore (unless you have very lowend needs or are extremely strapped for cash), but https://wikidevi.com/wiki/ZyXEL_NBG6617 might be an interesting option (https://github.com/openwrt/openwrt/pull/1098) Jun 24 22:14:04 pkgadd: 100mbps uplink, need wireguard or shadowsocks. Jun 24 22:14:21 koops: in that case ar71xx is definately out of the window Jun 24 22:15:33 pkgadd: Asus RT-AC58U? Jun 24 22:15:43 ar71xx can do 100 MBit/s, but it's right at the limit - and that's before you try anything special. SQM or VPN is definately too much for it Jun 24 22:16:20 I have an ASUS RT-AC58U. The factory installation method is a little messy, but you only have to do it once Jun 24 22:16:44 koops: https://wikidevi.com/wiki/ZyXEL_NBG6617 and RT-AC58U are pretty much the same, but the ZyXEL has twice the RAM (and that makes a difference) Jun 24 22:16:59 The only thing I'm not happy about is the RAM, which is only 128 MB Jun 24 22:18:11 It's pushing 18 days of uptime now, and the wifi firmware hasn't crashed yet Jun 24 22:18:47 There was something bad about Qualcomm IPQ806x too Jun 24 22:19:10 So basically no Qualcomm, no Broadcom, no MediaTek. Jun 24 22:19:23 That's going to exclude most options. Jun 24 22:19:38 pkgadd: thanks for Zyxel. Jun 24 22:19:43 The RT-AC58U is IPQ40XX, not IPQ806X Jun 24 22:19:51 ipq806x is a good choice, it's just not as fast in terms of WAN-to-LAN routing speed as mvebu or mt7621, so I wouldn't recommend it for WAN speeds above 300 MBit/s - the CPU is /very/ fast though Jun 24 22:19:52 I mean the other. Jun 24 22:20:04 AC2600 for example. Jun 24 22:20:15 Didn't Qualcomm abandon this SoC line? Jun 24 22:20:20 I think firmware crashes are common on units with external wifi radios, like the QCA9880 Jun 24 22:20:34 no upstream work on it Jun 24 22:20:36 well, that's true - and kind of an exaggeration at the same time Jun 24 22:21:42 So I can get one without worrying about support? Jun 24 22:22:04 if you have a 100 MBit/s WAN connection and the budget, both the Netgear r7800 and the ZyXEL NBG6817 would be very good option. but the ZyXEL NBG6617 should be a pretty decent alternative for about half the money as well Jun 24 22:22:46 RAM size is the only thing which would make me avoid the RT-AC58U Jun 24 22:23:36 you just have to know that nbg6617 support hasn't been merged yet (as the pull request has only been filed 2 days ago) Jun 24 22:23:42 MediaTek MT7621AT has issues Jun 24 22:23:56 two things running at the same time = lock Jun 24 22:24:04 and wi-fi ones I think Jun 24 22:24:40 Marvell also has Wi-Fi issues. Jun 24 22:24:54 Is basically every router crap? Jun 24 22:25:06 something will always have issues Jun 24 22:25:13 there is no perfection Jun 24 22:25:39 If I don't care about wi-fi and ok with wired, is Broadcom ok? Jun 24 22:26:09 no Jun 24 22:27:45 pkgadd: will it run very slow or something? Jun 24 22:27:57 if you're concerned about future maintenance of ipq8065, which is a popular highend router SOC, you should be tenfold concerned about broadcom Jun 24 22:28:14 lol Jun 24 22:30:14 pkgadd: Why are there no upstream commits for ipq806x? Jun 24 22:30:24 ask QCA Jun 24 22:30:51 they did submit them initially, but didn't follow up with them when they weren't merged as-is Jun 24 22:31:19 but at the same time QCA did provide the ipq806x kernel 4.14 kernel bump for OpenWrt Jun 24 22:31:57 so it will run? Jun 24 22:32:22 I want a router that fits OpenWRT like a glove. No random crashes etc. Jun 24 22:32:47 I'm using the ZyXEL NBG6817 for a bit over a year now - and I'm very happy Jun 24 22:32:52 Are these mediatek issues serious or is it exaggeration? Jun 24 22:32:54 cool. Jun 24 22:33:05 Do I have to build my own OpenWRT for this router? Jun 24 22:33:23 You said something about a PR that isn't merged. Jun 24 22:33:41 no, but you need 17.01.5 (not released yet) or >=18.06-rc1 Jun 24 22:34:06 snapshot? Jun 24 22:34:28 the PR was for the nbg6617, that one would require you to build your own/ patched OpenWrt /until/ it's merged in master - so for a few weeks Jun 24 22:34:35 snapshots are also fine Jun 24 22:34:51 nbg6617 == ipq4018, nbg6718 == ipq8065 Jun 24 22:35:13 nbg6617 ~80 EUR, nbg6817 ~170 EUR Jun 24 22:36:42 Linksys WRT3200ACM Jun 24 22:36:42 Netgear R7800 Jun 24 22:36:42 ZyXEL NBG6817 (Armor Z2) Jun 24 22:36:42 Linksys WRT1900AC Jun 24 22:36:42 Linksys WRT1900ACS Jun 24 22:36:42 Linksys EA8500 Jun 24 22:36:42 TP-Link Archer C2600 Jun 24 22:36:43 Netgear R7500 Jun 24 22:36:43 Linksys WRT1900AC Jun 24 22:36:44 ZBT WG2626 Jun 24 22:36:44 ZBT WE1326 Jun 24 22:36:45 D-Link DIR-860L Jun 24 22:37:00 use a pastebin for more than 2-3 lines Jun 24 22:37:05 ah. ok Jun 24 22:37:22 https://www.irccloud.com/pastebin/C3ISrLEU/ Jun 24 22:37:34 if 100% - it is much more understandable https://s33.postimg.cc/4ghykxppb/Image_3.gif Jun 24 22:37:35 so this? Choose any you want? Jun 24 22:38:00 Or there are some I shouldn't choose? Jun 24 22:38:20 I removed the ath ones. Jun 24 22:38:26 I wouldn't buy WG2626 (WG3526 is a better choice) or DIR-860L anymore, both use the older mt7602e 2.4 GHz wlan chipset, mt7603e (as in wg3526 or WE1326) is the newer option Jun 24 22:38:55 so does newer chipset fix locks when running two tasks? Jun 24 22:39:25 it's supposed to deal better with congested environments/ interference in the 2.4 GHz band Jun 24 22:39:30 than mt7602e Jun 24 22:40:15 so basically I need the one that can push gigabit LAN and 100Mb/s Shadowsocks/WireGuard Jun 24 22:41:34 all of your list can do that without VPN, if you want VPN towards those speeds, mt7621 probably won't cope (I'd tend towards ipq8065 then) Jun 24 22:42:17 (or ipq8064, if you find a /very/ cheap offer - but not for ~30-40 EUR less, then ipq8065 is a much better choice) Jun 24 22:43:41 I currently run Shadowsocks with chacha20-poly1305. Will mt7621 run it well? Jun 24 22:43:57 It's less computationally intense than AES. Jun 24 22:44:42 880 MHz mips vs 1.7 GHz ARMv7 Jun 24 22:45:30 mips=weak? Jun 24 22:45:59 mips is slowly disappearing from the highend router market Jun 24 22:46:20 Also worse OpenWRT support? Or not? Jun 24 22:46:26 no Jun 24 22:48:22 https://www.irccloud.com/pastebin/2nzkbML4/ Jun 24 22:49:21 Will MIMO and beamforming work on OpenWRT? Jun 24 22:50:22 drop the r7500 from your list, r7500v1 has unsupported wlan - only r7500v2 is supported Jun 24 22:50:29 should work Jun 24 22:52:02 Why isn't a list like this shown in this channel's topic? Jun 24 22:52:19 "Recommended high-end routers for OpenWRT" Jun 24 22:54:14 if "money isn't an issue", the top three of that list should all be good, the others can only compete on a price basis Jun 24 22:55:02 https://ufile.io/a6ltt Jun 24 22:55:21 not saying they'd be bad, not at all Jun 24 22:55:35 4096 MB flash on ZyXEL?! Jun 24 22:55:42 yes Jun 24 22:55:44 humor this doos with tcp pakcege Jun 24 22:56:41 4 MB spi-nor, containing bootloader, firmware, etc., plus 4 GB eMMC containining a dualboot set (4 MB kernel + 64 MB rootfs/ overlay) Jun 24 22:57:08 https://github.com/pkgadd/nbg6817/ Jun 24 23:02:28 I think it's wrong. There is no load on the processor, and there are no answers. Jun 24 23:06:36 qwefytuiityty: What are you trying to acheive? Jun 24 23:10:42 0.2 msec time interval pakceges and 0%, not 78% in lede/openwrt if load cpu 1% Jun 24 23:12:02 if it is real to do Jun 24 23:13:42 other hops have 0% Jun 24 23:14:03 pkgadd: I found TP-Link Archer C2600 used for 80 Euros. Jun 24 23:14:33 Get or not? Jun 24 23:15:06 If it's used I think it's better to reflash it completely including bootloader. Jun 24 23:15:43 If I get a router from US and will replace the power brick that ships with it with EU one, will it work? Jun 24 23:16:07 10.75.8.1 is local a gateway Jun 24 23:16:43 ather the same i think geatway/router Jun 24 23:16:48 other Jun 24 23:16:56 koops: sounds like a sensible, one thing you need to be aware of, TP-Link has made the serial console almost unusuable (check the wiki for details) - so if you want the option of serial console access, this wouldn't be the best option Jun 24 23:17:10 How am I going to reflash it then? Jun 24 23:17:19 Why did TP-Link do that? Jun 24 23:17:42 because they don't want to give you serial access, that's a very intentional and deliberate choice on their end Jun 24 23:18:29 others the same, i think gateway/router Jun 24 23:18:46 wtf why Jun 24 23:20:18 night Jun 24 23:20:33 finally Jun 24 23:20:43 that was a very large part of the reason why I didn't go with a c2600 myself (the others were flash size and the slightly older wlan chipset, all for just 30 EUR less than nbg6817/ r7800) Jun 24 23:21:03 why do they disable serial Jun 24 23:21:39 their understanding of FCC/ ETSI no-tampering requirements probably Jun 24 23:22:18 TP-Link has started playing that game for a while already, sometimes even combined with signed firmware upgrades on newer devices Jun 24 23:22:29 wasn't this cancelled? Jun 24 23:22:58 well, TP-Link still increasingly moves into that direction Jun 24 23:23:04 or was this model released when it was in effect Jun 24 23:24:49 https://www.irccloud.com/pastebin/RSaOwaAn/ Jun 24 23:25:41 mine was signed apparently Jun 24 23:25:55 or simply because it was RU Jun 24 23:26:08 which is more likely? Jun 24 23:26:26 regdom is pretty strict in .ru Jun 24 23:29:39 This is why I want to buy a router overseas. Jun 24 23:29:52 No risk of Cheburashka backdoor. Jun 24 23:30:22 I want to use a router for VPN after all. Jun 24 23:30:35 Don't want them to be reading all my traffic. Jun 24 23:31:20 for regdom reasons, I'd recommend a european source over a US one though Jun 24 23:31:30 If I buy used, how can I flash everything including u-boot? Jun 24 23:31:48 completely zero everything out and then write legit firmware. Jun 24 23:32:16 Same if my package looks like it's tampered with. Better to discard it in this case though. Jun 24 23:34:31 technically you have access to the complete flash (mtd-rw) once your have OpenWrt installed and ssh access, but you can't reflash 'everything' Jun 24 23:34:57 ART is device specific and you can't replace it with one from a different device Jun 24 23:35:54 Ask someone to dump it? Jun 24 23:36:29 it must fit your device Jun 24 23:36:58 yes, you could replace it, but it requires the calibration offsets for hardware variances of /your/ device, not someone else's Jun 24 23:44:19 that's not just necessary for regdom requirements, but also to get a stable/ clear signal Jun 24 23:46:24 man, really need someone to leak the docs under nda for qualcomms newer stuff Jun 24 23:50:08 hardware variances? Jun 24 23:52:14 manufacturing tolerances; for something like wlan is matters. very early during the nbg6817 development (that has been solved long time ago), parts of these (board-2.bin) weren't properly taken into account, with the result that I couldn't even connect to the AP before reducing txpower by 4 dB Jun 24 23:52:16 que? Jun 24 23:52:17 o Jun 24 23:54:33 if I will set antenna power to like 10W, will cops come for me Jun 24 23:54:50 the signal was visible and strong, but no client could connect/ succeed to auth. proper calibration data (ART) for /your/ device is important Jun 24 23:55:55 so if this data is infected you can't do anything? Jun 24 23:56:14 a) probably, yes (you're close to the cell phone and radar frequencies, so the proper licensees will notice and make sure that someone unfriendly knocks on your door) b) the hardware can't transmit at 10 wats without bursting into flames Jun 24 23:56:48 you can't really 'infect' that data Jun 24 23:57:23 what about corrupting it in a way that triggers a bug in a parser? Jun 24 23:57:46 by keeping the device working, very unlikely Jun 24 23:58:31 https://github.com/pepe2k/ar9300_eeprom/blob/master/dump/wdr3600_v1.1_2050500272_rev1.3_pl.txt that's more or less a (formatted) interpretation of the data contained in ART Jun 24 23:58:31 Can a shellcode fit in this data? Jun 24 23:59:11 64 KB, fit, yes, being usable - very unlikely Jun 25 00:00:26 What about a shellcode that downloads more data from the internet? Jun 25 00:12:03 pkgadd: Does ART contain some variable-length data that can trigger a buffer overflow? Jun 25 00:18:21 check the header files in the kernel Jun 25 00:32:32 build #56 of gemini/generic is complete: Failure [failed images] Build details are at http://phase1.builds.lede-project.org/builders/gemini%2Fgeneric/builds/56 blamelist: Matthias Schiffer Jun 25 00:46:20 pkgadd: Was there something wrong with ath10k? Jun 25 00:48:15 or Qualcomm Atheros QCA9984 Jun 25 00:48:41 no wlan driver is perfect, but it works for me Jun 25 01:14:07 Paranoia overload Jun 25 01:23:43 I tell you, my HDD is about to get me, with all the read/ write and smart errors it throws at me. may it rest in tiny pieces. Jun 25 01:24:13 Spinning rust? Jun 25 01:24:14 In 2018? Jun 25 01:25:24 SSDs are just too small (yes, / is on a SSD, but I still need space) Jun 25 01:25:44 if you need space and enough left to pay your rent, disks it is :D Jun 25 01:26:23 consumer 4tb ssds are still nearly 1000GBP Jun 25 01:26:27 heh Jun 25 01:29:19 128 GB on a SSD is good enough (well, at today's prices 256 GB), but everything else goes in spinning rust Jun 25 01:29:28 s/in/on/ Jun 25 01:29:48 yeah Jun 25 01:30:01 I have 256G, 128 is a bit tight Jun 25 01:30:10 and the price difference isn't that much Jun 25 01:31:22 HDDs are jstu too slow Jun 25 01:31:56 7200 rpm drives aren't that bad, 5400 rpm on the other hand really are Jun 25 01:32:03 Oh, 7200s are Jun 25 01:32:33 I've used a couple 10ks, still too slow Jun 25 01:33:08 Seek times are the bane of my existence Jun 25 01:33:16 doesn't really matter though, you use mechanical disks for storage, not quick access Jun 25 01:33:30 Exactly Jun 25 01:33:32 I need access. Jun 25 01:33:33 :) Jun 25 01:33:42 thats why you have ssd caches Jun 25 01:33:43 :D Jun 25 01:33:49 Nah. Jun 25 01:33:58 access times on the bigger modern disks aren't that bad either tbf Jun 25 01:33:59 If you want to play the cache game, just get more RAM Jun 25 01:34:25 well 512G ssds are almost affordable, reasonable enough as caches Jun 25 01:34:30 with a sensible filesystem, of course Jun 25 01:36:31 but depending on your application, you could do simpler things **** ENDING LOGGING AT Mon Jun 25 03:00:03 2018