**** BEGIN LOGGING AT Thu Mar 04 02:59:56 2021 Mar 04 08:23:10 guidosarducci: I find that exceedingly unlikely, tried to build master directly, without any of my patches. Something's fishy… :/ Mar 04 08:51:23 so i just cleared out some old files on my overlay, including passwd file, and notice new file has a user for ubus Mar 04 08:51:45 unfortunately, when a ubus user exists, there's some issue where procd cannot connect to ubus Mar 04 08:51:51 and the boot hangs Mar 04 08:52:13 it was running as root before with no issues, but when there's a ubus user and it runs as that, no joy Mar 04 08:52:14 any ideas? Mar 04 08:56:00 nyt: Have you updated ubus through opkg, at any point in time? Mar 04 08:56:11 neg, i think i found it Mar 04 08:56:14 might be old etc/config/rpcd Mar 04 08:56:21 gonna try clearing that out and see what happens, but not sure itllh elp Mar 04 08:58:29 https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=de7ca7dafadfd650d031e0379ce0c002868d5936 Mar 04 08:58:55 There was an issue, but it was fixed a while ago. Mar 04 08:59:07 fresh build off master Mar 04 08:59:09 doesnt seem related to that Mar 04 09:06:34 yeah so only change, if i edit passwd and remove the ubus line, so ubusd runs as root, everything works great Mar 04 09:07:33 otherwise, [ 18.729220] procd: Connection to ubus failed Mar 04 09:15:35 guidosarducci: Meanwhile, I set the patch as not applicable. I have no idea what's going on. Could it be a config (OpenWrt and kernel) problem? Mar 04 09:59:42 figured it out Mar 04 09:59:47 somehow passwd was umask 77 Mar 04 09:59:53 ubus would hang repeatedly trying to read it Mar 04 10:01:15 mangix: Hm… musl 1.2.2 is a bit chubbier. Mar 04 10:05:05 silly users demanding that previously skipped features be implemented :) Mar 04 10:07:33 karlp: I only complain about unnecessary chubbiness. :P Mar 04 10:07:57 And musl 1.2.2 had me at mallocng. :P Mar 04 10:14:07 (And 64-bit time_t, of course, but that's basic necessity.) **** BEGIN LOGGING AT Thu Mar 04 10:48:20 2021 Mar 04 10:52:54 build #15 of ath79/generic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-21.02/images/builders/ath79%2Fgeneric/builds/15 Mar 04 11:52:04 Hi, I need some help with porting OpenWRT to a device; a netcomm router (BCM6368) that has a mini PCI-e slot with a 3G modem inside. I don't know what it is, but after running binwalk on both the original and the builder's generated image, the partition order is wrong (something else?)? I'm not too sure what I need to change in the dts file/device config block; original firmware binwalk: https://i.imgur.com/Vk5KGUJ.png and the Mar 04 11:52:04 generated image binwalk: https://i.imgur.com/0CAj5Px.png Mar 04 11:52:31 I tried to install it on the device, but since the order is wrong, it refuses to upgrade to my image Mar 04 11:55:03 I've read through the docs too Mar 04 12:07:41 Festivenari: what's the flash layout the OEM firmware reports? Mar 04 12:07:50 pastebin instead of screenshot please Mar 04 12:07:58 Borromini: uh, sure. You mean /proc/mtd, right? Mar 04 12:08:01 does your openwrt initramfs boot? Mar 04 12:08:24 Festivenari: yes, but also what's reported in the boot log (both show the same but in a slightly different way) Mar 04 12:08:36 sure, let me get the device and extract that info Mar 04 12:10:37 svanheule: been looking for any openwrt supported outdoor APs but it really looks like that EAP225 Outdoor is the only one that has real PoE afaict, not the 'passive' PoE like some other stuff, and a gigabit uplink. Mar 04 12:10:48 slim pickings :P Mar 04 12:12:27 Borromini: https://pastebin.com/NVgv2N7X Mar 04 12:12:46 It's not a 4/32MB device technically (5MB flash), but it does have 2x USB ports Mar 04 12:15:13 And here are my source code modifications (dts file contents and bcm63xx.mk edits): https://pastebin.com/4W8amexV Mar 04 12:15:44 weird... the CHIP_ID is incorrect Mar 04 12:15:48 guess my IDE made a mistake Mar 04 12:16:18 but when I compiled it, it was correct Mar 04 12:18:16 oh, right... the boot log Mar 04 12:18:43 Festivenari: that first paste isn't showing the boot log Mar 04 12:19:07 Borromini: so, uh, would dmesg suffice? My serial to USB adaptor broke and I'm still waiting on them to come Mar 04 12:19:10 Borromini: Unifi UAP-AC-M also claims to support standard 802.3af http://dl.ui.com/datasheets/unifi/UniFi_AC_Mesh_DS.pdf Mar 04 12:19:41 Borromini: dmesg log: https://pastebin.com/a1mHHCY5 Mar 04 12:19:57 Borromini: not sure if there's any benefit over the 225-Outdoor (like a Mediatek radio) Mar 04 12:20:01 I could try and use my RPI to get the serial output, but hmm Mar 04 12:20:06 last I tried, it just wouldn't work Mar 04 12:20:44 I'm willing to sacrifice a few of these routers since I have so many on hand (which is the reason why I'm porting OpenWRT) Mar 04 12:21:20 svanheule: no, and with this MT7613 i don't know if having mt76 wireless is such an advantage xD Mar 04 12:21:32 :P Mar 04 12:22:48 specs look almost identical Mar 04 12:25:16 Borromini: actually, I will try and get the RPI for the entire boot log Mar 04 12:25:19 *use the Mar 04 12:25:31 Festivenari: that's your way to get serial on the device? Mar 04 12:26:15 Borromini: yep, until those adaptors arrive Mar 04 12:27:43 Borromini: UAP-AC-M is also ath10k https://fccid.io/SWX-UAPACM/Internal-Photos/Internal-Photos-3092372, so the TP-Link will most likely be cheaper Mar 04 12:28:15 Borromini: it does look like serial access on the UAP-AC-M might be easier than on the EAP225-Outdoor Mar 04 12:28:48 i.e. an actual header instead of PCB test points Mar 04 12:30:41 that's a pre yes. tp-link is 75-ish euros Mar 04 12:31:06 * Borromini hates Ubiquiti's ungoogleable product names Mar 04 12:31:08 oops... disconnected the wrong ethernet cable Mar 04 12:31:24 well UBNT is a tenner more expensive Mar 04 12:31:29 not the end of the world Mar 04 12:32:55 Borromini: /proc/mtd doesn't show addresses, just sizes. for the flash layout, the bootlog is really the only choice Mar 04 12:33:50 russell--: yeah that's why i asked both. Mar 04 12:34:00 /proc/mtd is easier for sizes :) Mar 04 12:34:13 /proc/mtd is useless Mar 04 12:34:41 well my hex isn't as fluent as yours no doubt. Mar 04 12:35:58 "dmesg | grep 0x0" does a decent job Mar 04 12:39:34 oh, great, I need a RS-232 level converter now... Mar 04 12:42:43 Borromini: well, is there another way of getting the boot log? Mar 04 12:42:59 can you ssh? Mar 04 12:43:06 russell--: yes, I have telnet access Mar 04 12:43:12 I can run any command as root Mar 04 12:43:17 Festivenari_: you really want serial access once you start breaking things. Mar 04 12:43:25 dmesg gives you the kernel part (post boot loader) Mar 04 12:44:16 Borromini: yeah, too bad the shipping time out of China sucks... Mar 04 12:44:37 Festivenari_: you can't get those things from local webshops? Mar 04 12:44:59 Borromini: uh, I can, but I mean, waiting 3 weeks or so isn't too bad, considering local eBay sellers would charge $10 each Mar 04 12:45:46 i'm not seeing any storage defined in your dts either btw Mar 04 12:45:55 but i'm not overly familiar with it Mar 04 12:46:07 Borromini: hmm... I took it from another device that had similar specs Mar 04 12:46:08 i buy 10-30 of the $2 dollar ones from china at a time to compensate for the slow shipping, then on average it only takes a few days each Mar 04 12:47:19 russell--: yeah, I suppose that's a very good idea considering it's too easy to short one of them out Mar 04 12:47:39 I might see if there's a store in my city that sells those adaptors Mar 04 12:49:26 Festivenari_: what part of the world are you in? Mar 04 12:49:38 russell--: Australia Mar 04 12:50:25 I already know the local places to get them, but I have a feeling those adaptors will come soon because it's been 1-2 weeks so far, so waiting a few more days can't hurt Mar 04 12:50:58 actually, might just order 1-2 of them from a local store, I suppose Mar 04 12:51:38 >$22.45 per adaptor Mar 04 12:51:41 yeah, think I'll wait Mar 04 12:53:37 :) Mar 04 12:55:24 Borromini: well, I'll probably return when they come. Well... thanks for the advice, I suppose Mar 04 15:28:24 mangix: ping Mar 04 16:45:05 anybody knows how to "fix" packages pull-request ci - are the fails only from too many parallel builds or sth similar ? Mar 04 18:27:13 Borromini: So today I discovered update_kernel.sh… :P Mar 04 18:27:36 Ah, he's gone. Mar 04 18:32:31 Borromini: Hey, there! :) Mar 04 18:38:26 rsalvaterra: hi! Mar 04 18:38:28 what's up Mar 04 18:58:13 I think I have loss of openwrt stopping issuing ip addresses to clien for second time.. By dnsmasq I think. 19.07.7 on x86. Is there a reason. This didn't happen on 19.07.6. Anything known? Mar 04 18:59:01 yawn Mar 04 19:01:54 Borromini: I discovered update_kernel.sh… :P Mar 04 19:02:41 rsalvaterra: welcome in 2017 :P Mar 04 19:03:13 So. Anyone want to hazard a guess as to why dnsmasq refuses to work unless the WAN port is connected and up? Mar 04 19:03:28 stintel: Hey, I'm not working on OpenWrt since 2017… XD Mar 04 19:03:37 rsalvaterra: how come? :D Mar 04 19:03:40 rsalvaterra: me neither ;) Mar 04 19:03:45 very handy huh ;) Mar 04 19:03:59 Grommish: bind-interfaces? Mar 04 19:04:32 And listening at the wrong "door"? Mar 04 19:05:17 Borromini: Indeed, but going from 5.10.18 to 5.10.20 is going to require manual work… :) Mar 04 19:06:41 lots of patches or just a few? Mar 04 19:07:13 i usually wait till graysky pushes his updates, he's always pretty quick bumping kernels :) Mar 04 19:07:36 Borromini: 5 reverse-appliable, 17 refreshed, as of yet. Mar 04 19:07:40 rsalvaterra: No.. If I leave the WAN port unplugged, dnsmasq on the LAN ports refuses to work.. it won't even bring the ports fully up.. they report up but th br-lan refuses to acknowlege even a ping Mar 04 19:07:51 rsalvaterra: the minute I plug WAN in and it pulls an IP, the rest works fine Mar 04 19:07:56 And one isn't applying at all. Some Freescale .dts change. Mar 04 19:08:21 rsalvaterra: and br-lan is statically set Mar 04 19:09:01 Grommish: How's the router connected? Mar 04 19:09:05 anything connected to the LAN ports before wan is up pulls a 169.x autoconfig IP Mar 04 19:09:26 rsalvaterra: It's got 3 Eth ports.. eth0/1/2. No switch chip. br-lan is 1/2 bridged, wan is 0 Mar 04 19:09:47 br-lan is static, wan is dhcp/dhcpv6 Mar 04 19:10:02 rsalvaterra: if it's really reverse applied, those are easy Mar 04 19:10:03 and it works - as long as WAN is up and has an IP Mar 04 19:10:19 but no wan = no lan for whatever reason Mar 04 19:10:24 Borromini: Yep, just git rm. Mar 04 19:12:14 Maybe it's just a hardware issue, but makes no sense that way Mar 04 19:12:16 Grommish: That's indeed very odd. Unfortunately I don't have an answer… :/ Mar 04 19:12:32 rsalvaterra: Me either >_< Mar 04 19:14:58 I don't believe it's a hardware issue, though. Mar 04 19:46:03 Is it my impression, or is jffs2 in (very, very light) maintenance mode…? Mar 04 21:25:07 5.10 bump to 5.10.20 pull sent. Mar 04 21:26:41 rsalvaterra: pong Mar 04 21:27:08 mangix: Hi! Just to let you know musl 1.2.2 is working like a charm here. ;) Mar 04 21:27:37 Wouldn't doubt it. Been running it for several months as well. Mar 04 21:27:57 Is it just me, or do I suddenly have more memory available? Mar 04 21:29:34 No idea. Maybe the malloc changes are to blame Mar 04 21:29:41 The used memory used to hover at about 62 MiB on my Archer C6, now it doesn't go above 56 MiB… Mar 04 21:29:59 Might just be placebo, but I'm happy anyway. :P Mar 04 21:30:40 I do believe the libselinux patch should be part of the series, though… Mar 04 21:34:12 Sure. I'll add it later Mar 04 21:36:59 There's an interesting bug with musl and PPC. libgcc has long double as 128 bits whereas musl has it at 64. No idea how to fix GCC. Mar 04 21:38:14 GCC9 and above I believe. Mar 04 21:39:34 Just to make sure, this is the patch you have, right? https://github.com/rsalvaterra/openwrt/commit/cbc2b627346affc1dcd581ecceb3bb4c0c014dad Mar 04 21:40:46 Yep Mar 04 21:41:11 Upstream fixed the issue. Will be part of 3.2 releasing soon Mar 04 21:41:24 It's at rc3 I think Mar 04 21:41:45 Well, if it's released before musl is merged, we won't need the patch anymore. :) Mar 04 21:43:59 I'm not hopeful it will be soon. Mar 04 21:59:19 mangix: Nope, it's definitely not placebo. Both the Archer C6 and the TL-WDR3600 are using less memory. Mar 04 22:10:17 rsalvaterra: that's...good i guess Mar 04 22:10:45 any theory why? Mar 04 22:15:42 mangix: mallocng. Less fragmentation. Mar 04 22:23:49 Grommish: would you mind if i bothered you when i'd start looking at a dedicated dts for the EdgeRouter Lite? Mar 04 22:25:42 zx2c4: ping. i am still carrying this patch in my own tree and wondered if there had been any further discussion on this since you asked florian if he'd like to see that in wireguard-tools proper? https://patchwork.ozlabs.org/project/openwrt/patch/20200512110634.21976-1-fe@dev.tdt.de/ Mar 04 23:05:36 Borromini: Not at all, although damex is probably the best for dts help. Do you have one already? or is there one in the ERLit uboot (if it has one)? Mar 04 23:06:21 Borromini: Or, you can pull it from the running system and we can see what's different Mar 04 23:07:02 Grommish: it's using the generic one that comes with the linux kernel atm Mar 04 23:08:13 Borromini: If you opkg install dtc, you can then run: dtc -I fs -O dts /sys/firmware/devicetree/base Mar 04 23:09:05 Borromini: and it'll export the running dts to stdout Mar 04 23:09:18 that's at least a good start Mar 04 23:09:50 Grommish: it's already supported on openwrt, just using the 'generic' dts that is upstream in the linux kernel Mar 04 23:10:32 Borromini: Yes, but if you generate a new one, we can compare them Mar 04 23:10:50 and see what needs to be changed.. or, is there a dts already for thge lite? Mar 04 23:10:54 floating aroudn Mar 04 23:11:16 ah, i see Mar 04 23:11:18 i don't think so Mar 04 23:11:20 sec Mar 04 23:11:29 Borromini: https://github.com/Embedfire/ebf-buster-linux/blob/master/arch/mips/boot/dts/cavium-octeon/ubnt_e100.dts Mar 04 23:11:37 Check that out :) Mar 04 23:11:53 but that looks like the internal kernel one Mar 04 23:12:09 so if that is what Owrt is using, it should find it. What needs to be added/removed? Mar 04 23:13:19 I'm still trying to fix my wsl install, so I can't dig into the source right now Mar 04 23:13:44 Grommish: i remember talking to damex about it, i thought it was on github. i'll have to crawl through my history Mar 04 23:13:55 no worries, the erlite is a testing device here, it's all good Mar 04 23:13:58 why wsl? Mar 04 23:14:03 it's my edgerouter 4 that's in production Mar 04 23:14:23 mangix: My main build system is a bare-metal install on another machine Mar 04 23:14:33 mangix: wsl2 lets me do it here as well Mar 04 23:14:55 But, that other machine is also my streaming/recording/editing box Mar 04 23:15:02 so, I like having more options Mar 04 23:15:21 i personally use wsl to ssh into my main box Mar 04 23:15:22 I give the Ubuntu WSL vm 16gb ram and it seems happy Mar 04 23:15:40 this laptop is too slow with its 4 cores Mar 04 23:15:54 but, either docker screwed it up, or my fstab changes did, or something else Mar 04 23:16:39 My other system is a Alienware M17xr4, so it's older, but I have 32Gb ram in it, a tb 7200RPM drive for windwo and a 500 ssd for Linux.. i7-3xxxx series Mar 04 23:16:56 THis one is a i7-7500 with 32gb ram, but no SSD, just a 2tb hhd Mar 04 23:17:10 They compile about the same though honestly Mar 04 23:18:01 so now I'm converting my wsl2 to wsl1 to see if I cna get it to boot, fix the issue, then convert it back.. I exported the image and its' 75gb on disk, so its going to take a while I suspect Mar 04 23:18:25 uhhh intel? Mar 04 23:18:44 AMD is much faster with its ridiculous L3 cache Mar 04 23:18:50 NOW it is ;p Mar 04 23:19:06 In 2013 when my AW was made, ot 2018 when I got the second one, it wasn't Mar 04 23:19:07 that L3 cache massively helps with compilationm Mar 04 23:19:35 :P Mar 04 23:20:09 i'm just waiting for Dell to release an XPS with Zen3. Mar 04 23:20:28 Grommish: i held out till zen to upgrade my ailing Phenom II X6 Mar 04 23:20:43 I want enough money for this: https://www.newegg.com/supermicro-mbd-h11dsi-n702-ma015-o-dual-amd-epyc-7000-series/p/N82E16813183691 Mar 04 23:21:38 Althought I hear they are going to release a new CPU SKU that has 96 cored Mar 04 23:21:44 err cores per CPU Mar 04 23:21:48 Grommish: https://youtu.be/la0_2Kmrr1E?t=181 Mar 04 23:22:04 Grommish: the rumored upcoming epyc you mean? Mar 04 23:22:11 * Borromini forgot the name Mar 04 23:22:32 It's another Italian name, yes Mar 04 23:22:40 rome or milan Mar 04 23:22:50 i don't know which one is the long awaited one :P Mar 04 23:23:04 Genoa Mar 04 23:23:17 oh! Mar 04 23:23:18 yes. Mar 04 23:23:20 see... Mar 04 23:23:21 >_> Mar 04 23:23:22 Genoa is the new one with 96 cores suposedly Mar 04 23:23:27 and i go to italy every year :P Mar 04 23:23:30 I had to look it up ;p Mar 04 23:23:32 well, went. Mar 04 23:23:38 ty covid >_> Mar 04 23:24:07 But yes, that is very tasty and way outside my price point Mar 04 23:24:14 but, we can dream Mar 04 23:24:16 hehe Mar 04 23:24:28 i'd be happy with a 5800X to bump from my 1800X atm :^) Mar 04 23:24:31 Especially since AMD is supporting ECC RAM out of the box with it Mar 04 23:24:35 yeah Mar 04 23:24:42 about time Mar 04 23:24:51 got pointed to this https://arstechnica.com/gadgets/2021/03/windows-com-bitsquatting-hack-can-wreak-unknown-havoc-on-pcs/ Mar 04 23:25:02 well AMD has been doing that before zen3 i think already Mar 04 23:25:16 Not officially Mar 04 23:25:27 and it's sportatic for folk to get it o wokr from what I've heard Mar 04 23:25:28 no, true Mar 04 23:26:24 Gah.. the system lags and bogs and my spelling and typing go to pot Mar 04 23:26:55 (that's my excuse and I"m sticking to it) Mar 04 23:28:12 =) Mar 04 23:30:03 I feel like I"m chasing my tail with this.. If you search for "Windows Subsystem for Linux has terminated unexpectedly" the "reasons" are all over the place ;/ Mar 04 23:30:17 Shocker, I tell ya.. but when it was working, it was brillant Mar 04 23:30:42 :P Mar 04 23:30:49 and I can't get it to tell me a reason ;p even with the custom event logger template Mar 04 23:32:48 Ah well, I will bookmark that dts file and see if it's in the kernel or not.. if not, I'll toss it into the tree Mar 04 23:34:10 =) Mar 04 23:34:22 i can't find the conversation i had with damex about it, unfortunately Mar 04 23:34:27 he did mention some things were missing Mar 04 23:45:43 that dts uses the cn3xxx dtsi to stub against so I moight have to grab both, though I think the dtsi is in the kernel already Mar 04 23:48:55 yeah Mar 04 23:59:55 i gotta go, way past bedtime again :( Mar 04 23:59:56 bbl Mar 05 01:16:00 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_lantiq.html has been updated. (98.2% images and 98.3% packages reproducible in our current test framework.) **** ENDING LOGGING AT Fri Mar 05 02:59:57 2021