**** BEGIN LOGGING AT Wed Dec 02 02:59:57 2020 Dec 02 03:00:17 dorf: email gstrauss@gluelogic.com Dec 02 03:00:37 mangix: he doesn't hang around here then? Dec 02 03:00:42 no Dec 02 03:00:58 shame on him! Dec 02 03:01:36 any idea who is(n't) maintaining i2pd, mangix? Dec 02 03:02:03 that one is nobody Dec 02 03:02:27 yeah, looks like it's not been looked at for a while. :| Dec 02 03:03:41 in other news, it seems there are no 11ax m2/pcie modules for sale anywhere Dec 02 03:03:42 I'm in regular communication with the i2pd lead dev.. if there was a simple make script for the opkg, maybe I could persuade him to integrate it into the CI. Dec 02 03:27:29 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_ar71xx.html has been updated. (99.2% images and 97.5% packages reproducible in our current test framework.) Dec 02 09:14:33 ynezz: thanks, that could be helpful, I'll try having a look after work Dec 02 12:49:19 if I'm seeing high RxFcsErr and RxAlignErr counters on an ar8236 switch, what should I be looking at? Dec 02 12:52:06 gardening Dec 02 12:54:43 * ldir- spurts tea Dec 02 12:55:43 😂 Dec 02 12:56:34 this is on TP-Link OC200 on the port connected to the CPU. the OEM GPL code is weird, it sets mvneta to phy-mode MII which isn't supported Dec 02 12:57:11 anyway, I tried doing everything that the OEM driver does, and I'm still getting packet loss, due to those 2 reasons Dec 02 12:57:26 and OEM firmware does not show this behavior Dec 02 12:59:18 maybe there's some extra tagging or something on the frames sent from the cpu to the switch? Dec 02 12:59:28 fcs error basically means bad checksum, right? Dec 02 13:00:03 different clocks? Dec 02 13:04:39 here's what I took from the OEM driver Dec 02 13:04:39 https://git.openwrt.org/773a32f2 Dec 02 13:05:53 and here's what OEM does: https://gist.github.com/497f86066cf654aca444ccb77b4ba3cc Dec 02 13:07:27 but this is for DSA, so maybe they're indeed enabling something QCA specific that is expected for DSA but causes problems when not using DSA? Dec 02 13:09:18 I recall something about a QCA specific tag in some register Dec 02 13:09:30 yeah, that would explain the misalignment and fcs error Dec 02 13:09:38 due to the extra leading tag Dec 02 13:11:41 ok, thanks for the hint, it might actually be very helpful Dec 02 13:16:43 bit 11 in port control register is what I should look at then Dec 02 13:36:40 seems to be disabled though :( Dec 02 14:49:09 Sigh. When you just want to do a simple make kernel_oldconfig CONFIG_TARGET=subtarget and the build system says "f**k you, I'm building the whole toolchain"… Dec 02 14:49:13 FML… Dec 02 16:00:01 that sort of thing typically happens when I've just done a bad update, didn't remember to take a copy of the known good firmware and now have to wait for the entire lot to rebuild. Dec 02 16:04:51 rsalvaterra: that removal of all CONFIG_HAVE_* and almost all CONFIG_ARCH_* looks like a bug to me Dec 02 16:05:19 ... in your mt7621 patch Dec 02 16:05:54 can you justify this by something I have overlooked, or is this new config-filter just not working correctly? Dec 02 16:07:32 adrianschmutzler: Those are almost always selected by the specific architecture. And in practice, I'm running my Redmi AC2100 with those changes. Dec 02 16:08:29 yes, having them selected by the architecture typically results in having them in the target config Dec 02 16:09:02 so, I'm wondering how that would have changed within that same kernel so drastically Dec 02 16:09:25 ich werde mir das mal genauer ansehen Dec 02 16:10:30 Yeah, I've been thinking the same, but the fact is I can't explain (yet, at least) how it works. :/ Dec 02 16:11:03 I will play around a little. sorry for German in between, was reading something else for a second Dec 02 16:11:08 Thanks, but I had to Google Translate that… :P Dec 02 16:11:46 By the way, thanks for fixing my blunder in lantiq… that was stupid, sorry. :( Dec 02 16:12:05 Completely missed the operator. Dec 02 16:12:52 well, i could have seen it as well, or tested, or wondered why it was actually duplicate for _all_ of them ... Dec 02 16:13:59 okay, building toolchain now ........ shouldn't have switched to 19.07 before Dec 02 16:14:23 I know that pain too well. :) Dec 02 16:26:50 hmm so upstream does indeed enable that QCA header mode on the switch CPU port; I am not. will try enabling it but I suspect it'll break things even further Dec 02 16:26:57 unless this mvneta driver somehow expects it? Dec 02 17:02:42 yeah as expected, with the header enabled it's completely broken Dec 02 17:03:05 I guess I'll have to dig deeper in the changes to the mvneta driver Dec 02 17:03:50 anyone has a link to armada 3720 ethernet datasheet handy? Dec 02 17:12:26 rsalvaterra: as I expected, this commit breaks it: https://github.com/openwrt/openwrt/commit/a8fb12a7d62c94fd1bbda332752e0f8c794f3702 Dec 02 17:25:30 adrianschmutzler: Wow. How does it work here at all…? O_o Dec 02 17:26:59 I just reverted that commit and then I get something that looks reasonable after "make kernel_oldconfig" Dec 02 17:27:20 Unless I'm totally mistaken somewhere Dec 02 17:28:04 Ok, I guess I need to dig deeper. Thanks for the heads-up! Dec 02 17:33:46 I suspect it's just some quoting issue, and the stuff in config-filter is not matched as a whole, but only it's fragments. But I don't really understand that commit in detail Dec 02 17:33:55 it's->its Dec 02 18:14:33 120 packets transmitted, 120 received, 0% packet loss, time 121146ms Dec 02 18:14:34 yay Dec 02 18:17:01 new_ctrl2 |= MVNETA_GMAC2_PORT_RGMII; -> new_ctrl2 &= ~MVNETA_GMAC2_PORT_RGMII; Dec 02 18:57:47 git master is failing: ERROR: package/feeds/packages/docker-ce failed to build. Dec 02 18:58:23 Looks like it is for PKG_SOURCE_VERSION stuff. Dec 02 19:14:29 ERROR: Expected 'PKG_SOURCE_VERSION:=dc9208a3303feef5b3839f4323d9beb36df0a9dd' in '../runc/Makefile', found 'PKG_SOURCE_VERSION:=ff819c7e9184c13b7c2607fe6c30ae19403a7aff' Dec 02 19:33:17 adrianschmutzler: So, um… do I need to do anything? :) Dec 02 19:43:55 Found the issue on github. https://github.com/openwrt/packages/issues/13998 Dec 02 19:47:43 rsalvaterra: not necessarily. Dec 02 19:48:24 I only have a rough understanding of what Felix's change does, so I'm not able to fully judge this thing at the moment. Dec 02 19:49:08 Well, an easy test is to check the .config.set before and after my patch. Dec 02 19:49:10 So, my current plan is to just wait until somebody will have a closer look (which might be myself at some point where I have more time). Dec 02 19:49:26 There should be no changes whatsoever. Dec 02 19:50:24 And that was essentially Felix's final question in his first e-mail, where I don't have the time to investigate deeper at the moment Dec 02 19:51:43 Alright, I'm going to do a build here and see if there are any differences. I'll reply to the email with my findings. Dec 02 20:34:49 evening Dec 02 21:02:38 evening Dec 02 21:52:49 o/ Tapper **** ENDING LOGGING AT Thu Dec 03 02:59:57 2020