**** BEGIN LOGGING AT Thu Jun 07 03:00:04 2018 Jun 07 03:04:46 okay miniupnpd seems bugged always listens on wildcard ip Jun 07 03:05:06 tried to change the init script to use network_get_ipaddr instead of network_get_device. Jun 07 03:05:28 ip shows fine in generated config stills listens on wildcard address Jun 07 03:05:29 nice Jun 07 03:11:22 andrewmuck: i'm playing with WrtNode2R which is very similar (also MT7688AN + MCU, just that it got STM32 instead of the ATMega on the LinkIt) Jun 07 03:12:40 the default f/w has some issues but GPIO using libmraa worked well, putting lede on works but I can not compile for libmraa and I need GPIO control. Jun 07 03:13:05 I also want to get a second ethernet port running as well, so I am motivated to gete lede to work for me. Jun 07 03:14:12 i took mine out and set it up, i had a really old build on it: Linux LEDE 4.4.70 #0 Tue May 30 12:34:09 2017 mips GNU/Linux Jun 07 03:14:14 andrewmuck: for GPIO you may need to configure pinctrl, take a look at other ramips device-tree DTS files to see how that works Jun 07 03:16:48 andrewmuck: why can't you compile that? Jun 07 03:23:19 compiling libmraa fails, looks like there is a version compatability issue I am trying to pin down Jun 07 03:23:41 dangole: will look at pinctrl Jun 07 03:24:05 if you want to get help with the compilation error, you'll need to pastebin it Jun 07 03:25:55 is there a reason this doesn't work? https://downloads.lede-project.org/snapshots/packages/mipsel_24kc/packages/libmraa_0.9.0-049ba5fa9f2d18ac0ec6729c46916b34998d3c5f_mipsel_24kc.ipk Jun 07 03:26:13 oops. i mean https://downloads.openwrt.org/snapshots/packages/mips_24kc/packages/libmraa_0.9.0-049ba5fa9f2d18ac0ec6729c46916b34998d3c5f_mips_24kc.ipk Jun 07 03:42:08 hmm.. i came up with this to isolate the router interface networks between the zones: https://pastebin.com/Vqc6VvXF Jun 07 03:42:34 10.0.1.254 is the router ip of the isolated zone Jun 07 04:05:11 but that doesnt work. regardless of rules order. to make an output rule only dest can be used? Jun 07 04:11:32 m4t: does the spi_cs1 to talk with the MCU over SPI work on the LinkIt Duo? Jun 07 04:11:51 m4t: i'm trying to get that to work on the WrtNode2R and all i can find is that they are using a modified blob Jun 07 04:12:11 not sure Jun 07 04:12:19 im trying to get libmraa working though... Jun 07 04:12:38 m4t: https://github.com/WRTnode/openwrt/blob/master/target/linux/ramips/files/drivers/mtd/ralink/ralink_spi_bbu.o Jun 07 04:13:23 m4t: MT7628/7688 SPI Driver By WRTnode Team ! Jun 07 04:13:57 yeah i dont think that is included. i don't see any spi module in the base image. Jun 07 04:14:57 m4t: they replace the spi-mt7621.c with their own blob which can work only for Ralink/MediaTek SDK Kernel 3.xx Jun 07 04:15:08 ok Jun 07 04:15:18 im not sure how the linkit is wired up, sorry Jun 07 04:15:24 it's probably on the seeed wiki Jun 07 04:15:25 @blogic: you mentioned the fact that SPI with CS1 works only half-duplex on MT7620/MT7628/MT7628. any other magic required to get CS1 slave device working, ie. MCU on LinkIt Duo? Jun 07 04:15:59 m4t: the linkit is similar to that wrtnode because it also uses SPI with CS1 to communicate with the MCU Jun 07 04:16:29 m4t: they just put an external reset as well (very smart and missing on the WrtNode, but easy to wire up with a single jumper wire on the base board) Jun 07 04:16:52 m4t: and on the LinkIt it's that ATMega MCU, on the WrtNode2 it's an STM32 Jun 07 04:20:00 wiki "If only dest is given, the rule matches outgoing traffic" i created a rule with only dest zone dest_port in luci its get listed as discard forward Jun 07 04:20:55 dangole: none Jun 07 04:21:05 well in half duplex mode you cannot read without sending\ Jun 07 04:21:20 you must always send at least 1 byte before you can read Jun 07 04:22:10 m4t: that is a spi/mtd/m25p80 all in one driver Jun 07 04:23:49 if there or will there be an announcement mailing list for openwrt? Jun 07 04:24:11 so regular users can get email when there is a new release ready to upgrade, or a security fix that mandates an upgrade? Jun 07 04:24:38 awww, there's no screen-bce in terminfo Jun 07 04:24:51 nobody is going to check openwrt.org daily for new releases Jun 07 04:25:34 platform detection in latest snapshot of libmraa seems broken Jun 07 04:25:59 it's returning UNKNOWN_PLATFORM (99) when calling mraa.getPlatformType() Jun 07 04:32:57 i think it's because /dev/mem isn't there Jun 07 04:33:37 are we stil okay to submit patches for 18.06? going to add the *-bce terminfo files, assuming it can get ticked off Jun 07 04:38:53 what should i be looking at if i'm not able to detect a SIM slot with modemmanager? Jun 07 04:39:25 would that kind of interface be pci or usb? Jun 07 04:51:11 shm0: luci is not aware of output rules Jun 07 04:51:18 shm0: it needs changes to recognize those Jun 07 04:52:03 ty jow i noticed in the firewall output it creates the proper output rule Jun 07 04:52:04 ausjke: no dedicated annoucnement list is planned. Announcements are usually made on the development mailing list Jun 07 04:52:19 and it seems like this is not an output rule problem Jun 07 04:58:59 i tought miniupnpd would directly send into the 'lan' network from isolated. but this is not the case. because i route ssdp between the networks. i had excluded the router ip in the forward ssdp rule. but in the in input isolated rule i didn't specify source ip addresses. so ssdp traffic from lan to isolated hits the miniupnpd there Jun 07 04:59:33 limiting ssdp on both input and forward seems to fixed it Jun 07 04:59:45 *to the isolated ip range Jun 07 05:00:57 but that still doesnt fix the problem that clients on the isolated network can reach the router interface in lan. i need to find some nicer solution for this Jun 07 05:02:18 * ldir wanders in Jun 07 05:03:34 * jow is already there Jun 07 05:05:15 build #871 of mpc85xx/generic is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/mpc85xx%2Fgeneric/builds/871 Jun 07 05:05:20 build #12 of omap/generic is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/omap%2Fgeneric/builds/12 Jun 07 05:16:32 build #11 of mpc85xx/generic is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/mpc85xx%2Fgeneric/builds/11 Jun 07 05:48:34 blogic: ping (zynq PR reminder) Jun 07 06:03:50 * ldir is amazed at how spectacularly different https://github.com/openwrt/openwrt/pull/1010/commits/e4b54ce87b9d2ef3a31779a3d60279c951c8dded and https://patchwork.ozlabs.org/patch/925838/ are that resolve into exactly the same thing. Jun 07 06:04:08 libmraa seems to work fine after setting CONFIG_KERNEL_DEVMEM=y and rmmod leds_gpio. got the wifi led toggling. Jun 07 06:04:10 ¯\_(ツ)_/¯ Jun 07 06:04:37 too bad dude left... :| Jun 07 06:12:16 hi people Jun 07 06:30:56 * ldir is really rather annoyed at bikeshedders Jun 07 06:31:44 sup Jun 07 06:32:06 fuck it Jun 07 06:36:42 :( Jun 07 06:36:59 ldir:don't let it screw it up your day; it's not worth it Jun 07 06:42:07 dedeckeh: it's good advice which I shall try to digest. I'm far too frustrated by what I perceive as willful misunderstanding to be able to constructively use your advice at the moment. Jun 07 06:42:39 ldir: whats the problem? Jun 07 06:47:29 jow: I'll get back in 6+ hours - need a break. Jun 07 06:53:53 jow:https://github.com/openwrt/openwrt/pull/1011 Jun 07 06:57:19 luaraneda: yarp Jun 07 07:00:13 blogic: thanks. The link just in case https://github.com/openwrt/openwrt/pull/942 Jun 07 07:17:38 wonder if there's any way to get these 3 into openwrt-18.06 before release... some waiting for review for "ages" ;) https://github.com/openwrt/openwrt/pull/937 https://github.com/openwrt/openwrt/pull/893 https://github.com/openwrt/openwrt/pull/993 Jun 07 07:21:38 nbd: https://github.com/openwrt/openwrt/pull/996 looks legit Jun 07 07:26:32 yep Jun 07 07:35:27 * blogic throws the alphabet at the github crowd Jun 07 07:36:19 :) Jun 07 07:50:43 i wish i had not started to review #1001 Jun 07 08:06:51 blogic it's a 4 meg device. Jun 07 08:07:19 I thorght that no more 4 meg devices were being aded to openwrt Jun 07 08:07:33 1001:21 GET https://assets-cdn.github.com/assets/frameworks-f6e6ce21346c0d2eb22def1e8534afcb.css 0 () Jun 07 08:07:36 useful chrome Jun 07 08:07:39 how about an actual error message Jun 07 08:16:34 Tapper: dont recall anyone ever saying so Jun 07 08:18:16 Will 18.xx even fit on a 4 meg device? Jun 07 08:18:36 with the kernel getting bigger eatch bump Jun 07 08:24:48 we are considering to build images with and without webui Jun 07 08:32:28 It's just a lot of time and effort for such crappy devices IMO but it's not my time or effort so..... Jun 07 08:33:17 At some point openwrt will have to stop building for 4 meg devices because they just will not work. Jun 07 08:34:15 It's simple for me to say that tho as I am not stuck with a 4 meg device Jun 07 08:35:21 what can you do with a 4 meg device today? no luci, right? so it'll basically do routing+NAT44+dhcp, and that's it? Jun 07 08:35:37 dumb AP! Jun 07 08:37:26 How hard would it be to do a dumb AP build with all the things removed apart from the stuff to run as a AP and luci? Jun 07 08:37:38 but what does OpenWrt bring to the table then, compared to the stock firmware? Jun 07 08:37:43 and how smal could you get it Jun 07 08:37:48 small* Jun 07 08:38:14 don't know there maby a safe kernel Jun 07 08:38:42 hm, right. Jun 07 08:39:35 air time stuff for wifi Jun 07 08:40:06 all the work dun on the wifi stack for buffer bloat Jun 07 08:41:52 there could even be a way of macking a build that is a dumb ap report back to the mane router running openwrt and show up on the mane luci page so the dumb ap would not have to have luci installed Jun 07 08:42:13 It could be a verry slim build then Jun 07 08:42:51 blogic, with and without luci would be nice even without considering 4 MiB devices imho Jun 07 08:43:10 yes Jun 07 08:43:51 Do you lot think there is any meret in what i say about dumb AP build? or is it just beond the scope of openwrt? Jun 07 08:44:03 beyond* Jun 07 08:44:20 why not Jun 07 08:45:29 like if you have 2 or 3 aps they could all show up in luci on your mane router. sounds verry cool to me. rite people i have to go my kids sports day. Jun 07 08:46:23 Tapper: I guess the limited hardware resources was the initial reason that today openwrt is very efficient in size and performance ? Jun 07 08:47:02 true that! Jun 07 08:48:26 A dumb ap build would be a good way of leting people reuse all the old routers that are not up to mane router duty Jun 07 08:55:51 <_lore_> ping nbd Jun 07 09:03:57 _lore_: pong Jun 07 09:06:00 blogic: mksquashfs4 doesn't support create big-endian filesystems? :o Jun 07 09:06:41 <_lore_> nbd: I understood the issue of page refcount yesterday Jun 07 09:08:06 <_lore_> the issue is due to sg_set_buf that uses virt_to_page instead of virt_to_head_page Jun 07 09:19:36 build #15 of cns3xxx/generic is complete: Failure [failed images] Build details are at http://release-builds.openwrt.org/18.06/images/builders/cns3xxx%2Fgeneric/builds/15 blamelist: Stijn Segers Jun 07 09:21:28 Tapper: why would I put openwrt on a dumb ap? Jun 07 09:21:45 usually the wireless performance and stability is worse Jun 07 09:22:17 _lore_: interesting Jun 07 09:22:29 jow: perhaps for a mesh setup? Jun 07 09:22:38 abenz: that is not "dumb ap" anymore :) Jun 07 09:22:44 :) Jun 07 09:22:47 there we already talk about advanced features again Jun 07 09:23:22 what people often tend to forget: openwrt fills a certain niche (advanced linux functionality on small soho device) Jun 07 09:23:52 if it suddenly ceases to work on low-end hardware and you require stuff like beefy arm or x86 anyway, you can as well switch to bsd or debian Jun 07 09:24:14 <_lore_> nbd: looking at dma code (on tx side) it assumes the packets in fraglist are not paged but it is not guaranteed in mac80211 code afaiu Jun 07 09:25:05 jow: I think this isn't a "oh, now openwrt will only work on beefy hw", more of a "let's skip the most anemic HW, and let's now require 8MB flash" Jun 07 09:25:33 how many 4MB flash devices have been sold in the past 5 years? Jun 07 09:25:46 probably more than you think Jun 07 09:26:02 that's why I'm asking. Jun 07 09:26:03 jow: that and better security. I wasn't worried much when I heard about the recent "vpnfilter" malware.. had I been running stock FW (based on a QSDK, which is based on an RC release of openwrt of a decade or so ago), then I'd be concerned Jun 07 09:27:41 most of those malwares use documented vulnerabilities that never get patched Jun 07 09:34:13 that reminds me Jun 07 09:35:08 my huawei fiber box from ISP, by default configured as an all-in-one. I set it to bridged mode and dial (pppoe) using lede which has various softwares for my needs Jun 07 09:35:27 the huawei keeps going back to normal router mode (disables bridge mode and activates wifi by itself) Jun 07 09:36:22 I disabled TR-069, ssh and telnet, but it still resets after it stays online for a few days Jun 07 09:37:15 I don't have something to sit between the fiber link so I sniff and see.. in the end I was advised to download the config file and modify it removing certain things.. and so far it has not reset :) Jun 07 09:44:41 wigyori, only supporting little endian was an intentional concession Phillip Lougher made to finally get squashfs merged mainline, without further bikeshedding Jun 07 09:50:10 pkgadd: ack, thx Jun 07 10:50:33 why is wpad-openssl now underneath wpad-mini in menuconfig? seems odd Jun 07 11:03:18 nbd: the latest mt76 update you pushed to master, does it have any fixes for 2.4ghz? (github main issue) Jun 07 11:03:35 I'm running 18.06 and if it does will jump to master Jun 07 11:03:57 or will it make it to 18.06 in a few days? Jun 07 11:04:18 thanks for your efforts Jun 07 11:06:50 karlp: yeah something seems wrong there, I got asked about including them on my non-wifi router configs Jun 07 11:06:54 probably shouldn't happen Jun 07 11:07:10 (these configs have nothing wpad/hostapd/wpa_supplicant enabled) Jun 07 11:07:12 "non-wifi"? Jun 07 11:07:35 yes, apu2, erl, ... Jun 07 11:08:28 not sure how that releates? I must have misunderstood something :) Jun 07 11:09:09 make oldconfig asks me about wpad-openssl when I don't even have wpad enabled Jun 07 11:09:17 doesn't seem right, so something seems wrong there Jun 07 11:09:27 nevermind Jun 07 11:09:45 ah, right, now I understand Jun 07 11:10:00 yes, someting seems, wrong, the whole tree of it changed, Jun 07 11:10:02 shouldn't multitask Jun 07 11:10:22 I find oldconfig to be a little difficult to see in context, it's just "here's this new option, Y/N/M?" Jun 07 11:18:13 wigyori: ok, context ? Jun 07 11:24:22 blogic: some netgear devices expect a bigendian squashfs i.e. the wnr612v2/n150r. for ath79, currently only squashfs4 is built, which doesn't support this. i had to add squashfs (3) into tools and use some of the old build functions Jun 07 11:24:33 which is not an issue, just some tweaks Jun 07 11:25:22 once i get the n150r running via ath79, i'll send some patches Jun 07 11:27:25 it's just that because only squashfs4 was added as tools for ath79, i thought the aim is to update the build functions to use that Jun 07 11:30:05 abenz: 2.4 ghz stability might improve with that update, but it needs more testing Jun 07 11:31:22 great. will flash and test. cheers Jun 07 11:42:25 jow: ping? Jun 07 11:54:20 which is he "future" again, block-mount or blockd? can we maybe update the package descriptions to help clarify this? Jun 07 12:01:42 if i'm having issues accessing a SIM slot on a devboard, is that going to be pci related, usb, or gpio? Jun 07 12:19:28 biangbiangmian: i think the sim card is directly wired up to whichever cellular module is on the board Jun 07 12:19:56 my modem goes into a mpcie slot Jun 07 12:22:22 yeah i think it uses specific pins on the mini pcie Jun 07 12:23:55 ok thank you Jun 07 12:24:03 my modem is showing up correctly though Jun 07 12:24:22 i'm using modemmanager and it is reporting that the sim is not found Jun 07 12:25:50 oh is the cellular modem part of the SoC or is it a separate module? Jun 07 12:26:22 seperate Jun 07 12:27:51 https://hackaday.com/2017/03/29/use-a-mini-pci-e-3g-card-with-usb-instead/ Jun 07 12:28:06 see how theyre wired Jun 07 12:29:10 ah, then i need to check the datasheet of my modem Jun 07 12:29:42 yeah it's confusing...usb in a mini pcie slot etc. Jun 07 12:30:16 maybe not as bad as m.2 though, which is both pcie, sata, and usb in the same slot Jun 07 12:30:59 they do have keys and notches to make this not as bad as all that... Jun 07 12:32:07 whatever happened to the one connector, one protocol paradigm Jun 07 12:32:49 https://xkcd.com/927/ Jun 07 12:35:04 build #16 of brcm47xx/mips74k is complete: Failure [failed images] Build details are at http://release-builds.openwrt.org/18.06/images/builders/brcm47xx%2Fmips74k/builds/16 blamelist: Stijn Segers Jun 07 12:35:48 m4t: it was dumb and results in thousands of connectors? Jun 07 12:36:02 this isn't new, in any way. Jun 07 12:36:09 isa and pci and agp all did the same things. Jun 07 12:36:41 what do you mean? Jun 07 12:37:48 m.2 is even worse; it can be either PCIe x2 or PCIe x4, but the slots are not forwards-compatible so an (m-keyed) x4 device won't work in a (b-keyed) x2 slot. There are also several other less common keyings. Jun 07 12:38:17 You can have cards with m- and b-keys at the same time, but those can only use x2. Jun 07 12:38:29 lol Jun 07 12:38:51 and now usb type-c is carrying analog audio Jun 07 12:39:09 and display port :) Jun 07 12:39:18 and power in both directions. Jun 07 12:40:02 hdmi does ethernet and audio return channel Jun 07 12:54:05 build #16 of brcm47xx/generic is complete: Failure [failed images] Build details are at http://release-builds.openwrt.org/18.06/images/builders/brcm47xx%2Fgeneric/builds/16 blamelist: Stijn Segers Jun 07 12:59:53 ^^^ drivers/usb/host/pci-quirks.c:543:6: error: redefinition of 'usb_amd_pt_check_port' bool usb_amd_pt_check_port(struct device *device, int port) Jun 07 13:01:49 jow: generic/pending/811-pci_disable_usb_common_quirks.patch Jun 07 13:02:00 jow: if you compare the version in master vs the one in openwrt-18.06 Jun 07 13:02:39 probably just sync that one from master to openwrt-18.06 and it should be good Jun 07 13:07:11 * ldir guesses someone else fell into that same trap as us stintel Jun 07 13:14:49 build #16 of apm821xx/sata is complete: Failure [failed images] Build details are at http://release-builds.openwrt.org/18.06/images/builders/apm821xx%2Fsata/builds/16 blamelist: Stijn Segers Jun 07 13:19:10 fixed it Jun 07 13:19:22 thanks :) Jun 07 13:19:37 tw Jun 07 13:19:39 yw& Jun 07 13:19:41 UGH! Jun 07 13:33:35 Hw do I use a .img instead of a .bin to go back to stock? Jun 07 13:48:04 802.11 fast roaming is for multiple APs on the same band right? not switching from 5 to 2.4 on the same AP? Jun 07 14:10:09 is this patch going to stick around? It breaks qos-scripts config parsing, I will have to either fix qos-scripts parsing or just revert the patch... https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=73d8a6ab7668173d70adbed45b61be5256c505e1 Jun 07 14:10:42 netprince_: it is going to stick Jun 07 14:10:53 jow: ok, thanks Jun 07 14:19:40 jow: thanks, will write a script to monitor downloads.openwrt.org and email when new releases coming out, need a way to know it automatically and email is still the ubiquitous tool Jun 07 14:20:19 https://github.com/openwrt/openwrt/pull/925 visiting all the 200+ PR requests this one is of the most interest, rbm33g is a very interesting board Jun 07 14:20:30 hope it can make into 1806 Jun 07 14:37:21 hi all. Jun 07 14:40:24 karlp, 802.11r works between 2.4 and 5 GHz just as well as between different APs, we just don't have bandsteering (so the client decides) Jun 07 14:42:16 please help me.. i'm build for mikrotik rb750 r2 (but with breed bootloader - not routerboot. i'm use ap143 as base) openwrt from sources and have problem - luci apply wrong tagged/untagged vlans to ports. in web luci menu i'm create some vlans - example - vlan 201 tagged on port 2 and vlan 202 untaged on port 2. In /etc/config/network - all good - ports - vlan 201 - '0t 2t' vlan202 - '0t 2'. Jun 07 14:42:26 I just went back to split SSIDs, too hard to diagnose what's going on when they're both the same ssid :) Jun 07 14:42:52 I may set up a second AP and try it again then, but can't endlessly screw around with the home wifi. Jun 07 14:45:12 But in swconfig show - i'm see that vlan wrongs (tag untag) - vlan 202 - 0t and vlan202 0t 2t Jun 07 14:45:35 if add only tag or untag - all works good Jun 07 14:46:08 problem only if port need tag and untag on same port.. Jun 07 14:46:38 where maybe problem? luci? swconfig? uci? Jun 07 14:54:20 sekil_: swconfig Jun 07 14:54:38 not all ICs support untagged + tagged on the same port Jun 07 14:54:48 and of those who do, not all swconfig drivers support it Jun 07 15:10:48 blogic: Who knew a shift register needed that much code? Jun 07 15:32:00 Monkeh: indeed Jun 07 15:32:31 wigyori: ouch Jun 07 15:32:43 wigyori: and we cant add a dummy squashfs3 hdr ? Jun 07 15:32:54 Monkeh: i'll need to look into that in more detail Jun 07 15:33:06 Monkeh: imho there is a shift register driver Jun 07 15:35:03 Any news on procd-ujail which is still broken on trunk ? https://bugs.openwrt.org/index.php?do=details&task_id=1559 Jun 07 15:35:43 blogic: There is - gpio-74x164 Jun 07 15:36:18 thats the one Jun 07 15:36:26 ledeuser: no Jun 07 15:36:53 ledeuser: did not see the bug :/ Jun 07 15:38:54 blogic: dunno, i'll check Jun 07 15:39:21 blogic: btw, give a monkey to whoever added the avm fritz300e support, as that also depends on squashfs3 :) Jun 07 15:39:34 wigyori: some AVM boxes require the same so we append a static 128 byte dummy squashfs3 hdr and then the real rootfs behind it Jun 07 15:39:47 yep avm fritz Jun 07 15:39:53 well, currently it's the append-squashfs-fakeroot-be foo, which uses mksquashfs-lzma Jun 07 15:40:02 target/linux/lantiq/image/eva.dummy.squashfs Jun 07 15:40:08 ok, Jun 07 15:40:14 try just adding that eva file Jun 07 15:40:24 and it'll spare us an additional tool Jun 07 15:40:27 :) Jun 07 15:40:30 will do, thx Jun 07 15:40:39 cool Jun 07 15:45:56 blogic: any ideas how to debug this procd-ujail issue, considering that it's intermittent (sometimes it works and sometimes it doesn't)? Jun 07 15:47:06 ledeuser: hm, considering that it randomly happens maybe look into the direction of available entropy / blocking reads on /dev/random etc. Jun 07 15:47:36 the instand reboot on killing ujail could be procd getting upset (and still being init) Jun 07 15:47:52 *instant Jun 07 15:48:15 maybe some procd crash due to unexpected service state or some other unexpected condition Jun 07 15:49:19 ledeuser: since you're able to obtain a shell in the hanging state, an "ls -lh /proc/$(pidof ujail)/fd" would be helpful to see what kind of descriptors are open Jun 07 15:49:34 ledeuser: strace -p $(pidof ujail) might reveal something too Jun 07 15:50:14 ledeuser: you could also try disabling the reboot on panic through sysfs and see if you get a kernel oops on killing the stuck processes Jun 07 15:50:56 not sure if that even works on openwrt, but these kinds of steps are what I would tryx Jun 07 15:51:08 * ldir tries kill -9 1 and is surprised the system stays up Jun 07 15:56:08 ldir: Pay no attention to the terminally confused. :) Jun 07 15:57:22 jow: thanks for the helpful ideas. I've just rebooted the VM (same one I used to file the bug report) 4 times, came up fine everytime! LOL :-) Jun 07 15:57:51 ledeuser: it knows you're looking at it. Jun 07 15:58:50 ldir: The infamous quantum bug Jun 07 15:59:44 what's the key changes between 1806 vs 1701, other than kernel upgrades and newer packages Jun 07 15:59:56 ausjke: the name :D Jun 07 16:00:21 Hi any one have the link to the 18.6 snapshots? Jun 07 16:00:23 so it's an updated release, which says the core framework is solid :) Jun 07 16:00:55 jow: hope you can get some time for luci2, or just luci+lua5.x Jun 07 16:01:26 I'll polish luci+lua for the time being and slowly modernize it where possible Jun 07 16:01:52 I gave up on the modern web stuff, it changes faster than I can devote time to read into it Jun 07 16:02:08 here's a preliminary detailled changelog https://openwrt.org/releases/18.06/changelog-18.06.0-rc1 Jun 07 16:02:16 will get updated once its done Jun 07 16:02:29 Modern web stuff -> overcomplicated slow nightmare Jun 07 16:02:36 that's true indeed, maybe web-component can help a bit for webui, but god knows Jun 07 16:02:48 aha, got it to hang with 2 ujail processes! Well, "ls -lh /proc/$(pidof ujail)/fd" isn't any different than the VM that booted normally 6 fds: 0-2 -> /dev/null, 3 -> anon_inode eventpoll, 4-5 pipe:[2819] Jun 07 16:03:04 ausjke: last time I looked react was stil lthe thing, and before that it was angular2, before that angular1 Jun 07 16:03:29 now I hear rumours about "shadow dom" and "styled components" and "transpilers" and "typescript" Jun 07 16:03:46 this scares me, I rather continue writting shitty html with JS sprinkled in Jun 07 16:04:17 its not fancy, its not 2018ish but it mostly works and is reasonably small Jun 07 16:04:25 jow: Get some fresh college grads to write a new UI - it'll only need about 64M flash to itself Jun 07 16:06:05 Monkeh: and come in a docker container! Jun 07 16:06:25 * mamarley twitches. Jun 07 16:06:32 the last guy who knew how this container was created is also long gone, but hey, you can install it with "curl | sudo sh" Jun 07 16:06:36 Hell that is one big changelog I didn't think there had bin that menny changes Jun 07 16:08:03 Tapper: ausjke: this change log is unfiltered in the sense that I didn't yet try to filter out commits that got cherry picked into 17.01.x already Jun 07 16:08:13 Over a year of active development, it adds up Jun 07 16:08:18 which is maybe a few dozen Jun 07 16:08:39 OK but it's still a lot of changes. Jun 07 16:08:47 Well dun you lot Jun 07 16:08:53 :-) Jun 07 16:09:21 are we at rc1 yet? Jun 07 16:09:51 I need to flash and kick the tyres Jun 07 16:21:13 jow: if use swconfig instead of luci - all work good - but when apply config in luci - vlan breaks... Jun 07 16:21:31 jow: agree with the webui stuff, simple html+js will always work, along with bootstrap4 that is Jun 07 16:21:31 sekil_: you mean swconfig on the cli ? Jun 07 16:21:36 yes Jun 07 16:21:48 sekil_: then swconfigs processing of the generated /etc/config/network is bad Jun 07 16:22:16 sekil_: whats the switch IC used on this board? Jun 07 16:22:25 maybe.. but swconfig parse /etc/config/network ? Jun 07 16:22:29 yes Jun 07 16:22:33 qca9533 Jun 07 16:22:38 builtin Jun 07 16:22:41 switch Jun 07 16:23:06 strange.. :) Jun 07 16:23:32 okay Jun 07 16:23:58 you mean that - luci use swconfig to apply new config? Jun 07 16:24:01 you could pastebin the working command sequence and the (broken) /etc/config/network generated by luci and I'll see if I can take a look later Jun 07 16:24:17 luci just writes /etc/config/network and instructs netifd to reload the network config Jun 07 16:24:43 i'm having an openwrt talk this Saturday for a local linux event and writing slides now Jun 07 16:25:08 this will eventually call swconfig load which directly parses the switch_vlan bits from /etc/config/network Jun 07 16:25:09 trying to get something 'new' to the slides, including something like, 1806 in one month! :) Jun 07 16:25:37 so it is not like luci invoking a batch of swconfig commands Jun 07 16:25:50 ok Jun 07 16:25:59 but /etc/config/network always good Jun 07 16:26:03 an "/etc/init.d/network reload" would likely have the same (broken) effect Jun 07 16:26:32 only in kernel? or where swconfig getting info about vlans - bad Jun 07 16:26:48 sekil_: then you maybe missed a step in luci? especially in older versions you must adjust the ifname entries manually after setting the vlans Jun 07 16:27:05 no.. ok Jun 07 16:27:11 e.g. when promoting a cpu port from untagged to tagged you must change lan ifname from eth0 to eth0.123 as well Jun 07 16:28:00 yes - i'm undesratnd all... problem only when add vlans from luci.. if add vlans from swconfig cli - all good. i'm send you all info... Jun 07 16:28:07 thanks :) Jun 07 16:28:26 whait me :) Jun 07 16:28:37 and by adding vlans with swconfig cli I assume you mean something like "swconfig dev switch0 vlan 1 set ports ..." Jun 07 16:28:55 and not something like "vi /etc/config/network; /etc/init.d/network restart" Jun 07 16:29:25 the latter is what luci does (simplified ofc) Jun 07 16:30:40 sekil_: I will need to run now, if I do not get around to it tonight I will look tomorrow. Feel free to pm me or ask back here again tomorrow Jun 07 16:33:39 stintel: sorry about messing up .48 Jun 07 16:33:52 either i copied it wrong from ldir or ldir messed up :^) Jun 07 16:33:55 * Borromini hides Jun 07 16:36:29 * ldir throws one of monkeh's bananas at Borromini .... pointy end first. ;-) Jun 07 16:36:40 jow: https://pastebin.com/bhmncTbq Jun 07 16:36:47 Oi oi oi, careful with my banana stock Jun 07 16:37:08 :'( Jun 07 16:37:16 at least peel them before throwing :P Jun 07 16:37:17 so banana equivalent dose is a thing Jun 07 16:37:22 for radiation Jun 07 16:37:24 so i can just open my mouth Jun 07 16:37:53 :-) Jun 07 16:38:58 jow: you can see that /etc/config/network different from swconfig show - from swconfig show - all vlans on port 4 - tagged... but from luci - vid 2 - need untagged Jun 07 16:39:06 * ldir thinks it's going to take a LOT of bananas to 'deal with' Borromini via the gift of radiation poisoning :-) Jun 07 16:39:15 hahaha Jun 07 16:39:21 putin, is that you? Jun 07 16:39:27 cuz my name ain't skripal Jun 07 16:39:37 yay Jun 07 16:39:44 RC RELEASES ON MY BIRTHDAY Jun 07 16:39:48 god i am blessed Jun 07 16:39:52 https://openwrt.org/releases/18.06/start#roadmap Jun 07 16:40:15 i didn't even know that page was hiding there Jun 07 16:40:20 for two weeks at least! Jun 07 16:40:25 that was novichok. litvinenko (sp?) was err, 'removed' via polonium poisoning. Jun 07 16:40:46 Borromini: fortunately you're not in England... so you're quite safe from Putin :-) Jun 07 16:40:47 i prefer 'taken out of the equation' Jun 07 16:41:03 ldir: russians don't know better than that belgium is the capital of brussels Jun 07 16:42:11 at least, the russians that matter, you know. just like the donald. Jun 07 16:58:24 jow: https://pastebin.com/bhmncTbq Jun 07 17:18:57 nbd: Ran the MT7628 client through tests again - performance issues vs MT7603 AP appear resolved with UDP traffic (updated that to latest, also), TCP TX still fails in the same circumstances (WDS on ath9k AP, non-WDS on MT7603 AP). Jun 07 17:25:17 well damn Jun 07 17:25:25 it seems i have hard bricked the asus wl500gP Jun 07 17:26:05 seems like the functionality design of openwrt made me do it >_< why wasnt wan by default made to be one port like in most commerical routers ._. Jun 07 17:29:34 Snowblind: what are you talking about? Jun 07 17:31:07 Snowblind: did you have a different nick earlier? what's the context? Jun 07 17:34:00 nope Jun 07 17:34:18 well after flashing openwrt the default functionality wasn't just what i expected Jun 07 17:34:25 but doesnt really matter now since the router is fully bricked Jun 07 17:35:07 so... if you connect to one of it's lan ports, you don't get a dhcp assigned address from 192.168.1.x range? Jun 07 17:35:16 (wifi is off by default) Jun 07 17:39:16 confused, openwrt does nothing unusual with its wan port Jun 07 17:40:04 And how exactly did it make you brick it? Jun 07 17:45:47 Monkeh maybe with cutting a trace after trying PIN9 reset Jun 07 17:45:59 * Monkeh blinks Jun 07 17:46:08 karlp not really, i would get the ISP's dhcp response and basically it passed through to the PC Jun 07 17:46:12 via the router Jun 07 17:46:16 acting as a dummy switch Jun 07 17:47:29 you cut an unknown trace and now have no idea why it might be broken? :D Jun 07 18:10:58 nbd: Turned the tables, 7603 as sta, 7628 as ap - same failure on the 7603 side. No problems with the 7603 connected to the ath9k ap so far. Jun 07 18:46:25 r7106 (kernel 4.14.48) lantiq xrx200, HH5a, www pages partially loads then WAN disconnects, can see IP in overview-doesn't stay connected for more than 90 seconds, previous version no WAN IP at all Jun 07 19:21:35 FYI, I just compiled latest Openwrt trunk x86_64, even changed to gcc 7.3 (from 5.5), sysupgrade two VMs with Jun 07 19:23:41 oops, sysupgraded the two VMs, but the procd+ujail issue remains (assuming that ujail is the problem and not something completely different) Jun 07 19:24:18 and killing the ujail process (not pid 1) causes immediate reboot of Openwrt. Jun 07 19:25:39 comparing ls -lh /proc//* between a working and a non-working system shows no real differences. Jun 07 19:27:04 Any ideas why "logread" would stop working on a system with ujail stuck ? dmesg still works but shows no errors. Jun 07 19:32:15 The 2 VMs have a rather "complex" setup (multiple VLANs, openvpn, strongswan, mwan3, ddns, dhcp etc) but it's a test configuration that I've been running for 2+ years and it worked fine with LEDE 17.01 ... Jun 07 20:34:49 ldir: ping Jun 07 20:35:57 so after some sleep. i came with this rule to prevent input/output leakage between zones: https://pastebin.com/nKipDinS. I couldnt get addrtype to work regardless of src-type or dst-type or using both. but pkttype seems to work fine. Jun 07 20:37:28 regarding cake and the arp problem on docsis. i removed the arp = cs7 reference in cake. also changed the sqm script to only mirror ip traffic to the ifb device. that brought the voice tin packet count down by alot. Jun 07 20:38:09 but im still having the issue that my latency raise by around 10-15ms when playing online. but im not sure what the cause is. either the router, my isp, or the game server itself? Jun 07 20:38:43 Hi schm0 are you on virgin UK? Jun 07 20:38:55 no Jun 07 20:39:01 OK Jun 07 20:39:45 do you think my isp is deprioritize some traffic over time? Jun 07 20:40:09 I have a docsis. modem and I have to set my MTU to 1492 to make sqm work at it's best Jun 07 20:40:30 i also readded back the mvebu tx workaround patch. i have the feeling it takes longer now for the latency to raise. but could also be placebo x) Jun 07 20:40:39 mtu of 1492 on docsis? Jun 07 20:41:17 With my MTU at 1500 my modem was fraging my packets Jun 07 20:41:26 hmm. Jun 07 20:41:42 1500 seems fine here without fragmentation Jun 07 20:41:53 k Jun 07 20:42:03 Must just be my ISP then Jun 07 20:42:23 its strange. like 20min latency is around 23ms then it starts raising slowly above 30ms almost to 40ms Jun 07 20:45:59 when even romania has better internet then you :| Jun 07 20:46:04 haha Jun 07 20:46:33 100mbit fiber for 12$ Jun 07 20:46:37 a month not bad Jun 07 20:46:55 I get 200 meg but have to pay £35 Jun 07 20:47:10 but coax/docsis? Jun 07 20:47:14 yeah Jun 07 20:47:26 dsl/docsis is a dead end Jun 07 20:47:49 fuck nos mate as long as it works! :-D Jun 07 20:48:00 xD Jun 07 20:48:13 fiber is a better investment in the longterm Jun 07 20:48:56 but yeah now they spend money for vectoring for the next years. until they will notice i can't be improvement anymore. then they will start to deploy fiber everywhere Jun 07 20:49:34 so they can keep the people busy and waste more money in a longer period of time Jun 07 20:49:44 they know their shit for sure Jun 07 20:50:55 one isp brought 1gbit docsis into one major city months ago. still doesn't work properly. but now they start in the city. Jun 07 20:51:02 amazing Jun 07 21:02:09 one major docsis isp has big problems with ingress. they had to move tv channels because of ingress from dab+ Jun 07 21:03:09 and everytime when there is thunderstorm alot of people loose internet connectivity Jun 07 21:03:12 x) Jun 07 21:03:50 https://imgur.com/a/FV0pvF3 Jun 07 21:05:48 i got one problem with miniupnpd miniupnpd[4073]: http://10.0.1.254:5000/rootDesc.xml not found, responding ERROR 404 Jun 07 21:08:46 i think the file should get generated "on-the-fly". Jun 07 22:22:03 iirc, miniupnpnd is archaic sw and hasn't been touched since it was written, it doesn't surprise me remotely that it could be better. Jun 07 22:23:52 Is OpenWRT (including the LEDE release) vulnerable to VPNFilter? Jun 07 22:23:55 it crashes when you send too many requests at moderate speed Jun 07 22:24:10 problem is that most people don't know the status for each package, they should be 'rated' somehow, so people chose the actively maintained alternatives instead Jun 07 22:24:46 that's farrrr out of scope of openwrt though. Jun 07 22:25:02 try searching for "comparison of quality of upnp daemons" in general Jun 07 22:25:12 this isn't something that could possibly be maintained just for openwrt. Jun 07 22:26:04 something like debian's popularity package statistics will be nice Jun 07 22:26:14 uhhh Jun 07 22:26:15 anyway do we have a name for 1806, or will it have one? Jun 07 22:26:20 miniupnpd still gets updated Jun 07 22:26:58 * ausjke is working on openwrt slides Jun 07 22:31:15 popcon says _nothing_ about whether things work or not. Jun 07 22:31:27 mangix: might have been one of the other barely documented upnp daemons. Jun 07 22:31:54 is miniupnpd still using string functions to generate xml rpc in C? Jun 07 22:36:58 build #16 of cns3xxx/generic is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/cns3xxx%2Fgeneric/builds/16 Jun 07 22:47:04 beats me Jun 07 22:47:27 karlp: i'm guessing you're confusing dd-wrt's upnp daemon Jun 07 22:49:31 maybe, miniupnpd does seem to be actie, even if it's limited in scope Jun 07 22:49:59 so much more to upnp than just opening ports for bit torrent clients :| Jun 07 22:54:43 indeed Jun 07 22:57:50 build #523 of ramips/mt76x8 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ramips%2Fmt76x8/builds/523 Jun 07 23:27:43 miniupnpd was updated a few weeks ago Jun 07 23:28:10 latency also increases without cake. so that can be ruled out Jun 07 23:31:51 but does someone know why target and interval differ on egress vs ingress when using cakes diffserv4 implementation? Jun 07 23:32:27 like here https://pastebin.com/PYT7RJpR Jun 08 00:06:08 Snowblind: I don't really know what you've done with your Asus WL-500gP v1, but mine is totally fine with current LEDE master/ HEAD, http://paste.debian.net/hidden/792316a6/ Jun 08 00:07:11 (yes, I did replace the bcm4318 wlan card with an Atheros AR2414 ages ago, but that doesn't have an impact on the switch configuration) Jun 08 00:12:01 hmm Jun 08 00:12:21 back to the multi queue ethernet thing.... Jun 08 00:13:58 on mvebu there are 8 queues each for eth1 and eth0. the eth1 ones get replaced by cake. however on eth1 when i invoke tc -s qdisc show.... only 1 queue seems to transmit queues. i guess because of the tx queue workaround patch? Jun 08 00:14:07 for wlan0/1 there are 4 queues Jun 08 00:14:22 but only 2 seem to transmit packets Jun 08 00:14:53 should i replace the main mq qdisc with fq_codel to get rid of those multi queues? Jun 08 00:18:39 if i remember correctly there was only 1 queue on every device some time ago Jun 08 00:24:41 and it uses mq as main qdisc to distribute other qdisc like fq_codel over the hardware queues. i can't find any useful info about mq? it is some sort of dummy qdisc. but i can't find in the kernel config? Jun 08 00:34:46 hmm the wifi queues are used for wmm i guess? Jun 08 01:41:12 build #17 of brcm47xx/mips74k is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/brcm47xx%2Fmips74k/builds/17 Jun 08 01:57:18 build #17 of brcm47xx/generic is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/brcm47xx%2Fgeneric/builds/17 Jun 08 02:19:32 build #17 of apm821xx/sata is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/apm821xx%2Fsata/builds/17 Jun 08 02:38:07 hmm Jun 08 02:39:28 wmm and cake Jun 08 02:40:22 wmm uses cs4 to cs5 for video. cake would put cs4-cs5 into the voice tin. Jun 08 02:41:44 so i should not use cs3 for video but rather af4x Jun 08 02:42:01 and cs7 should not be used cause its reserverd? **** ENDING LOGGING AT Fri Jun 08 03:00:18 2018