**** BEGIN LOGGING AT Tue Mar 23 02:59:56 2021 Mar 23 03:11:01 Grommish: make rules live (at least to some order of approximation) in Makefile, which says: Mar 23 03:11:03 clean: FORCE rm -rf $(BUILD_DIR) $(STAGING_DIR) $(BIN_DIR) $(OUTPUT_DIR)/packages/$(ARCH_PACKAGES) $(BUILD_LOG_DIR) $(TOPDIR)/staging_dir/packages Mar 23 03:11:41 Meh, so yeah, monolithic.. Thanks :) Mar 23 04:38:55 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_ar71xx.html has been updated. (99.1% images and 97.9% packages reproducible in our current test framework.) Mar 23 05:35:46 ugh Mar 23 05:35:56 can someone ban this linmob guy? Mar 23 06:41:07 Hi people, i would like to develop a modem-router for university. Do you have an idea on SoC company to take chips from? It's better if openwrt full supported. Mar 23 06:41:43 CloudVII: Broadcom doesn't opensource their drivers, keep that in mind Mar 23 06:42:22 CloudVII: I'd check the forum. https://forum.openwrt.org/c/hardware-questions-and-recommendations/13 Mar 23 06:43:05 Watching here: Mar 23 06:43:06 https://openwrt.org/docs/techref/hardware/soc Mar 23 06:43:21 MediaTek? Did you try using their stuff? Mar 23 06:43:32 Although most univ hardware will be already well managed Mar 23 06:44:58 When you are talking schools, especially Univ MANs, you're not dealing with Off the Shelf hardware typically Mar 23 06:45:35 I mean, a project for university. Not to be used inside the CAN. Mar 23 06:46:40 If you're not trying to justify getting a new router, go cheap.. ALthough, have you checked out GNS3? Mar 23 06:47:53 GNS3 will allow you to setup a lab, and you can use Openwrt images Mar 23 06:47:59 and its free Mar 23 06:48:30 I'm an ee student, choosing project and thought about router. Could start with Turris Omnia since it's open. Asking if someone here worked with chips from xdsl vendors. Mar 23 06:48:32 Check out VyOS as well Mar 23 06:48:51 MediaTek seems the most open Mar 23 06:48:54 Are quite closed, they ask for reliable company, NDA. Mar 23 06:49:39 Wifi drivers seem to the alot of issues. Mar 23 06:49:56 But, that forum link I sent earlier will help you address those questions Mar 23 06:50:09 Then you can scan the ToH for the chipset type Mar 23 06:50:49 I mean, hope to have almost quite ready once done with the PCB design. Used openwrt on my home router years ago and worked perfectly. Mar 23 06:51:40 *nod* You're looking to design a router around OpenWrt if I understand correctly Mar 23 06:52:10 Yeah, because you did a great work. Don't want to lose a lot of time on software part. Mar 23 06:52:49 CloudVII: https://forum.openwrt.org/c/hardware-questions-and-recommendations/13 is where you should look. They know which hardware to suggest.. Although I'm surprised soeone hasn't in the past Mar 23 06:53:18 Someone did fortunately. Mar 23 06:53:20 https://project.turris.cz/en/hardware-documentation Mar 23 06:55:16 CloudVII: https://forum.openwrt.org/t/poll-interest-in-an-openwrt-purpose-built-router/69295/4 Mar 23 06:55:50 Someone asked the question you are asking, it was a few months ago, but might help Mar 23 06:56:27 There is always th RPi4b :D Mar 23 07:00:25 if it needs wifi, i'd probably go for a chip with enough pci-e slots and put some cheap intel wifi ax cards in there Mar 23 07:01:27 If he's deigning the pcb, just build it internal Mar 23 07:01:41 and pick the chipset for wifi Mar 23 07:02:18 says it's for an ee project, so is designing the pcb Mar 23 07:02:24 i actually wouldn't. i'd love a cheap board with 2-4x lanports, and 2-3 pci-e m.2 slots. that way i can put there what i want Mar 23 07:02:38 true Mar 23 07:03:05 can someone ban this linmob guy? Mar 23 07:03:11 I've got a device that I need to find someone to mount a 52-pin mPCIe header to the pads Mar 23 07:27:55 CloudVII, vdsl or cable modem part? Mar 23 07:28:15 vdsl only has lantiq working now afaik Mar 23 07:28:40 xdsl, vdsl Mar 23 07:29:49 Sorry, VDSL it's part of xDSL family. So xDSL. Mar 23 07:29:59 If you advice on xdsl chip you are welcome. Mar 23 07:30:04 *have Mar 23 07:35:30 CloudVII, as i said - lantiq modems are supported/running with vdsl Mar 23 07:36:16 https://xdarklight.github.io/lantiq-xdsl-firmware-info/ Mar 23 07:54:45 CloudVII, lantiq IP should now (2020) be MaxLinear (lantiq->Intel->maxlinear) Mar 23 08:08:51 infineon->lantiq->Intel->maxlinear Mar 23 08:09:18 ifx->ltq they payed 900m$ intel to maxlinear they pated 65m$ Mar 23 08:09:20 :-D Mar 23 08:12:41 intel has problems Mar 23 08:24:29 blogic: Nice work ucentral. I had just started working on my own small config management system, maybe time to revise ... :) Mar 23 08:43:03 kristrev: how did you find that ? Mar 23 08:43:12 kristrev: the new server code is almost ready Mar 23 08:43:31 I am aiming for an early PoC release next week and then a v1.0 in ~3 months Mar 23 08:43:35 blogic: I saw the ynezz starred it and got curious :) Mar 23 08:43:48 kristrev: http://ucentral.io/docs/ucentral-schema.html Mar 23 08:43:53 the data model is growing Mar 23 08:44:25 thanks Mar 23 08:44:52 I only took a quick look yesterday, but I hope to find some time to start experiemnting soon Mar 23 08:45:04 cool Mar 23 08:45:12 day job is pushing for something config management fast, but my solution is fortunately not written in stone Mar 23 08:45:29 it follows some of the same concepts, and thin lua layer on top of a model of the packages we have to support Mar 23 08:45:52 the cfg looks like this Mar 23 08:45:53 https://github.com/blogic/ucentral-config/blob/main/Home.venue Mar 23 08:46:10 we then render that out into uci after verifying it Mar 23 08:46:25 there will also be a progressive web app soonish to allow cfg foo Mar 23 08:46:50 nice Mar 23 08:46:56 and raspberry Pi images that allow you to run the controller locally without the need for it being in the public Mar 23 08:46:59 bah, wish I had seen this before, oh well Mar 23 08:47:14 kristrev: its fresh out the press Mar 23 08:47:35 there is a background healthcheck Mar 23 08:47:49 are there any plans to add functionality for exporting a configuration? Mar 23 08:48:19 so every N seconds the unit will test all the local dns/dhcp server, check if the wifis that are up are what uci says should be up, if the inet ifaces are operational, if the radius server is reachable (if using EAP) .... Mar 23 08:48:27 exporting in what way ? Mar 23 08:48:39 like we can add anything we want if there is a valid use case Mar 23 08:48:57 based on the healthcheck the unit then reports its sanity 0-100 Mar 23 08:48:59 with a web frontend it might not be so useful, but if I have already configured a device and would like to deploy that configuration to multiple devices Mar 23 08:49:13 ah yes Mar 23 08:49:22 then it would be practical if I could run a magic command that creates the model and import it on the other devices Mar 23 08:49:30 so the concept is that you do not cfg one device but a network of devices Mar 23 08:49:44 like you have a cfg and then say "deploy to the following devices" Mar 23 08:49:59 yup Mar 23 08:50:01 https://github.com/blogic/ucentral-config/blob/main/Home.venue#L45 Mar 23 08:50:12 if you look here, wifi is not cfg'ed based on a phy but a band Mar 23 08:50:23 so ucentral figures out by itself which phy to use Mar 23 08:51:02 what does the u and l mean? Mar 23 08:51:09 upper lower Mar 23 08:51:15 ah, thanks Mar 23 08:51:25 https://github.com/blogic/ucentral-schema/blob/main/cfg/cfg_phy.uc#L94 Mar 23 08:51:28 also stuff like this Mar 23 08:51:51 we check if mimo, htmode, channel will work on a phy, if not we dropp the the best effort version and notify the backend Mar 23 08:52:05 nice Mar 23 08:52:34 I see one issue with some of my devices, but it is probably not a big deal. I have multiple 2.4 or multiple 5 phys Mar 23 08:52:58 yeah, we can add a field to force a specific one Mar 23 08:53:11 aha, I see the function wifi_generate now, that answers my earlier question. Should have checked the code more closely before asking ... Mar 23 08:53:15 the unit will upon boot create a huge json with all the probed capabilities and send it to the backend Mar 23 08:53:35 so backend knows exactly whats in the ap Mar 23 08:53:36 https://github.com/blogic/ucentral-schema/blob/main/cfg/cfg_network.uc#L263 Mar 23 08:53:44 networks are created based on a type Mar 23 08:53:54 I am currently using the path in my cfg mgmt solution. My files are only valid for a device of a certain type, so it works for me (tm) Mar 23 08:53:59 so nat is like lan but without the eth ports Mar 23 08:54:10 guest is like nat but with isolate and local ip ranges blocked Mar 23 08:54:22 i'll be adding wireguard and vxln soonish Mar 23 08:54:26 *vxlan Mar 23 08:54:36 cool Mar 23 08:54:37 jow is currently working on the ipv6 support Mar 23 08:55:17 I look forward to following the progress and hopefully have time to dig deeper into the tool Mar 23 08:55:28 other things on the list are captive portal, hs2.0, open roaming, ... Mar 23 08:55:44 and the best part is, this is funded work ;) Mar 23 08:55:52 cool, can I ask by whom? Mar 23 08:55:55 so I am actually working on this almost full time right now Mar 23 08:56:18 (and I am sorry for not contributing usb to udevmand like I alked about earlier ... :( Mar 23 08:56:24 https://telecominfraproject.com/wifi/ Mar 23 08:56:32 hehe Mar 23 08:56:38 we'll add it eventually Mar 23 08:56:50 udevmand is actually used to push the topology map to the backend Mar 23 08:57:15 is it a continuation of purple or something else? Mar 23 08:57:30 its similar but in a different market Mar 23 08:57:42 it is not as ISP driven as prpl Mar 23 08:57:51 much more productive and technical Mar 23 08:57:59 prpl is more a manager meeting group Mar 23 08:58:00 sounds good Mar 23 08:58:09 TIP is far bigger aswell Mar 23 08:58:16 I just saw a lot of the same names on the list Mar 23 08:58:23 but I guess that is hard to avoid :) Mar 23 08:58:30 https://telecominfraproject.com/participants/ Mar 23 08:58:34 yeah Mar 23 08:58:49 there are like 600 members Mar 23 09:00:42 huge Mar 23 09:02:03 time for a meeting (yay), nice talking to you and again, nice work (to you and everyone else) Mar 23 09:03:17 blogic: friendly ping on this one… :) https://patchwork.ozlabs.org/project/openwrt/patch/20210316090236.2018-1-rsalvaterra@gmail.com/ Mar 23 09:03:39 (I'd really like to reduce our upstream diff as much as possible.) Mar 23 09:46:00 uco Mar 23 09:46:06 oops, sorry Mar 23 19:33:33 the buildbots are bored... Mar 23 19:38:27 Rather quiet in here today… :P Mar 23 19:49:14 Very Mar 23 19:51:39 Huh. What is this bmips thing? DTS version of bcm47xx? Mar 23 19:56:53 mangix: bcm6368, bcm6362, bcm6318, bcm6328 and bcm63268 Mar 23 19:56:59 (quoting noltari) Mar 23 20:01:33 OK Mar 23 20:23:05 mangix: Heh. I bit the bullet and did the 5.10 bump myself: https://github.com/openwrt/openwrt/pull/4023 Mar 23 20:23:27 It was actually easier than last time, since I didn't have to manually rebase any patches. Mar 23 20:33:12 rsalvaterra: at some point it usually gets 'easy', when the patchlevel creeps up Mar 23 20:45:40 Build [#19](https://buildbot.openwrt.org/master/images/#builders/2/builds/19) of `layerscape/armv7` failed. Mar 23 20:47:23 So is 18.06 EOL now? Mar 23 20:49:49 mangix: why? Mar 23 20:53:47 Just wondering. Mar 23 20:55:25 wasn't that announced? Mar 23 20:55:42 yes it was: Mar 23 20:55:44 "The current old-stable version series of OpenWrt is 18.06, with v18.06.9 being the last service release of the series. It was released on 9 December 2020. No new releases in this series are expected." Mar 23 20:55:50 from the front page. Mar 23 21:06:57 Heh Mar 23 21:09:55 aparcar[m]: sounds like the 18.06 branch should be locked Mar 23 21:13:43 rmilecki: trying to add support for a linksys device. The factory image has some header beginning with GOLF. Any idea about that? Mar 23 21:16:45 mangix: i don't recall any format using GOLD Mar 23 21:16:46 *GOLF Mar 23 21:17:04 mangix: provide me more bytes, please Mar 23 21:17:10 hexdump -C -n 1024 foo.img Mar 23 21:19:22 rmilecki: https://gist.github.com/neheb/a4c073998c854e8b760059ca5db3456f Mar 23 21:19:45 mangix: no, i don't recognize that :( Mar 23 21:20:18 I was able to generate a similar image with KERNEL := kernel-bin | lzma | fit lzma $$(KDIR)/image-$$(firstword $$(DEVICE_DTS)).dtb but I have no idea about this GOLF thing Mar 23 21:20:33 mangix: out of my power Mar 23 21:22:14 mangix: later part - that kernel@1 - looks familiar- it's u-boot and DTS Mar 23 21:22:24 scripts/mkits.sh - but you probably saw that Mar 23 21:22:40 aparcar[m]: Would like your input: what's involved in adding a build-system dependency and rolling out updates to buildbots, etc? Is it preferred/easier to add a host build that handles all that automatically? For my PR #3855, I add tools/elfutils to get libelf and libdw. Mar 23 21:23:10 mangix: you could try to look into the vendor upgrade application Mar 23 21:23:11 rmilecki: yeah I saw that. I think this is similar to the Linksys E8450 but I don't see any factory images Mar 23 21:23:36 open it with ghidra and then search for GOLF Mar 23 21:24:03 dango has some frankenstein instructions on the e8450: https://github.com/dangowrt/linksys-e8450-openwrt-installer Mar 23 21:35:29 hmm funny Mar 23 21:35:55 if I go in the bootloader directly, I don't need an image with a header Mar 23 21:36:14 unfortunately it gives me a lzma compressed: uncompress error 1 Mar 23 21:58:43 mangix: it's common that extra header is used by web UI only Mar 23 21:59:05 mangix: vendors often don't care to modify bootloader to understand their custom headers Mar 23 22:06:37 mangix: seems so. Mar 23 22:06:56 can't figure out what format this bootloader expects. bah. Mar 23 22:07:29 it seems to support both lzma and gzip Mar 23 22:16:47 guidosarducci: I'll look at it Mar 23 22:17:26 I can't find the PR Mar 23 22:17:49 noltari: do you have a device which requires https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=2024547e1f71ddbf94deecc5e2789d802f748377 ? if so, I'd be happy to help replace this with kmod-owl-loader Mar 23 22:51:22 anyone advise on a realtek device to buy for testing? Mar 23 22:51:28 ideally some kind of switch thin Mar 23 22:59:15 I bought both a Netgear GS108T and GS110TPP to play with, the latter with PoE+ - there's only a pad for serial, not sure if this is better from the other manufacturers... Mar 23 23:05:44 Nick_Lowe: thanks Mar 23 23:32:49 aparcar[m]: strange, PR #3855 is https://github.com/openwrt/openwrt/pull/3855 Mar 23 23:39:29 guidosarducci: strange indeed, I'll catch up on that in a bit and see what i can do Mar 23 23:43:49 aparcar[m]: thanks, opinions seem to differ on host tools vs system dependencies so I'd like to settle on a direction. Mar 24 00:28:52 aparcar[m]: the zyxel gs1900 family is also quite nice (models with- and without PoE), rather similar across the whole family and (so far) no h/w revision mess (D-Link in particular is shipping Marvell, Broadcom or Realtek, depending on the exact h/w rev and keeping the same model number) Mar 24 00:31:15 populated serial header on the board, 3.3V, 2.54mm spacing - 16 MB flash, but dual-firmware, so the effective image is a bit limited (6976k) **** ENDING LOGGING AT Wed Mar 24 02:59:57 2021