**** BEGIN LOGGING AT Tue Apr 20 02:59:56 2021 Apr 20 04:22:15 For devs using/maintaining ppc, was a QEMU-based target ever looked into, similar to armvirt or malta? Any foreseeable problems? A virtual ppc target would be handy for coverage and testing. Apr 20 04:48:42 guidosarducci: I don't think there are many PPC devices Apr 20 04:49:00 There are some but the vast majority is MIPS and ARM Apr 20 04:53:43 mangix: I know that, but ppc has a history of causing.. err revealing other SW problems, hence useful for test/debug. While it *is* possible to simply build one's SW against ppc targets (and I do), being able to run it too is better. Apr 20 05:00:58 big endian ppc as a whole is rather dead Apr 20 05:37:18 stintel, thank you ! Apr 20 08:30:39 plntyk: you're welcome! Apr 20 11:27:52 * rsalvaterra looks at the interrupt distribution on his MT7621 device and wonders… Apr 20 11:28:21 … has anyone around here tried irqbalance on MT7621? Apr 20 11:30:05 The distribution I see doesn't look bad (Ethernet on one core, both Wi-Fi devices on the other core, one on each sibling), though… Apr 20 11:34:35 yay got my DACs Apr 20 11:51:57 stintel: DACs? As in audio ones? Apr 20 11:52:10 no, Direct Attached Copper Apr 20 11:52:28 to link the CN6640-SNIC10E to my switch ;) Apr 20 11:53:21 for a DAC ... I'm waiting for Khadas Tone Board 2 in board/kit to become available Apr 20 11:53:32 I will then try adding it to my DIY UcD400 PA Apr 20 11:55:30 Overkill. Both of them. :P Apr 20 11:56:58 rsalvaterra, o/ Apr 20 11:57:22 nitroshift: Hi, there! Apr 20 11:58:01 hello! if i suspect a packaging bug in some things in the packages tree, would i discuss that here or in #openwrt? Apr 20 11:58:07 stintel: Actually… that UCD400 looks sweet. Single channel, class D… Apr 20 11:58:11 (of course i'm happy to file tickets or PRs but i first want to make sure i'm not missing things) Apr 20 11:59:00 rsalvaterra: the kit was dual channel Apr 20 11:59:30 rsalvaterra: wanted the Nc400 actually and got confused with the 400 # :P Apr 20 11:59:41 rsalvaterra: but the Nc400 is only available in single channel kit Apr 20 12:02:04 Habbie: if it's purely about packaging I'd say here is better - do mind that not all people who have access to the packages feed are on IRC Apr 20 12:02:25 is it about any of the PowerDNS daemons :) Apr 20 12:03:15 stintel: What preamp are you using? Apr 20 12:03:46 rsalvaterra: it's not really a pre, but I wanted to move my most expensive speakers away from the main amp to reduce the risk of clipping when I'm going nuts :P Apr 20 12:03:58 NAD T778 Apr 20 12:05:08 and the DIY kit was also kind of an experiment Apr 20 12:06:30 when I build my own place, I'll go nuts on object based surround with something like Anthem AVM 70 prepro, and each channel on some NcXXX monoblock Apr 20 12:07:11 with the monoblocks as close to the speaker as possible and balanced xlr from the prepro to the monoblocks Apr 20 12:07:36 or that's the idea that's currently in my head, at least :) Apr 20 12:10:10 stintel: I'm happy with my cheap-as-chips TPA3116D2, for the time being. XD Apr 20 12:10:59 that's all that matters! Apr 20 12:12:21 stintel, thanks! and yes, i'm currently looking at 'opkg install dnsdist' which needs 19.9MB and i suspect it could be smaller Apr 20 12:12:31 for example Apr 20 12:12:31 Besides, getting the lady to approve all the cabling… yeah, wound't be a trivial task. Apr 20 12:12:40 *wouldn't Apr 20 12:12:47 Habbie: ahh that's pretty awesome! smaller sounds good Apr 20 12:12:52 libh2o-evloop appears to contain 3 identical copies of a .so Apr 20 12:12:55 not symlinked, not hardlinked Apr 20 12:13:07 rsalvaterra: I'd probably hide the xlr in the walls and ceiling Apr 20 12:13:13 and ruby is pulled in for no reason that i can find Apr 20 12:13:20 (also boost but that's an empty shell) Apr 20 12:14:53 I'm actually running pdns-recursor on OpenWrt and I'm asking myself questions about the memory usage when loading large blocklists and using multiple threads Apr 20 12:15:17 i can probably answer those - might make more sense in OFTC #powerdns but i'm not picky Apr 20 12:15:37 alright, good to know (haven't paid much attention to it yet as it's some kind of side project) Apr 20 12:15:43 I have to run now, meeting Apr 20 12:20:17 actually I have both dnsdist and pdns-recursor on that device Apr 20 12:21:00 -rw-r--r-- 1 stijn users 1,3M dec 12 07:46 bin/packages/aarch64_cortex-a53/packages/dnsdist_1.5.1-1_aarch64_cortex-a53.ipk Apr 20 12:21:13 I guess you're talking about dnsdist + deps Apr 20 12:21:36 yep Apr 20 12:21:53 the dnsdist ipk is around 1,3-1,5M yes Apr 20 12:22:03 but with all deps it adds up to 19,9M installed Apr 20 12:22:35 alright, it's hard for me to check that as I'm building custom images Apr 20 12:22:44 ah Apr 20 12:22:56 i'm just toying with the rootfs image from the docker hub right now, and installing via opkg Apr 20 12:23:14 so opkg install dnsdist pulls in ruby? Apr 20 12:23:34 yes, because libh2o-evloop depends on it Apr 20 12:23:38 i think it is a build dep for it Apr 20 12:23:42 but not a runtime dep Apr 20 12:24:07 so that might be as simple as removing one word from libs/h2o/Makefile Apr 20 12:24:25 the package defines it as a normal dependency, not a build dep Apr 20 12:24:47 PKG_BUILD_DEPENDS:=ruby/host libwslay Apr 20 12:24:57 i guess -this- line means it needs ruby on the machine that is running the compiler? Apr 20 12:25:00 yeah, that's a host build Apr 20 12:25:10 and then later it indeed lists it as a normal dep, and i can't see a reason for that Apr 20 12:25:15 so it needs ruby on the host to be able to build it Apr 20 12:26:42 seems that dep is there since initial commit of h2o (c9fa74bf43cfa535d6220a9db8cf43e20c4f104c) Apr 20 12:27:21 don't know if the maintainer is on IRC Apr 20 12:27:29 he's been on #powerdns on occasion Apr 20 12:27:31 but not recently Apr 20 12:27:42 maybe drop him an email first to ask why that dep is there? Apr 20 12:27:47 yep, makes sense Apr 20 12:28:07 maybe it's just a leftover before he figured out PKG_BUILD_DEPENDS ruby/host Apr 20 12:28:22 from before* Apr 20 12:28:44 likely - the 3 copies of the .so also probably stem from inexperience (while he certainly has more than i do, at this point) Apr 20 12:28:57 i'll ping him while i get a build env set up Apr 20 12:28:59 thanks Apr 20 12:29:08 I can actually try and remove that dep Apr 20 12:29:18 I have a working build env Apr 20 12:29:29 'dan graag' :) Apr 20 12:29:42 geef me ff ;) Apr 20 12:51:58 nice, tools/zstd doesn't build Apr 20 12:52:18 always the same problem when you want to "quickly do X" Apr 20 12:52:21 hehe Apr 20 12:52:49 looks like zstd itself has not been touched on master for a while Apr 20 12:53:06 maybe I need a make clean Apr 20 12:54:16 it looks like i am a bunch of makes away from being able to do anything, but that's to be expected as a new user i'd say :) Apr 20 12:54:49 it's building a lot of things my Debian already has ;) Apr 20 12:54:52 yeah the problem is that the device that I'm running pdns-recursor/dnsdist on is not supported yet and my current work doesn't rebase cleanly on master Apr 20 12:54:56 ouch Apr 20 12:55:04 i'm keeping life simple on x86 for now Apr 20 12:55:41 what device is it? Apr 20 12:55:53 I've got arm, arm64, mips, mips64, powerpc64, x86 Apr 20 12:55:58 this is a TP-Link OC200 Apr 20 12:56:26 PoE-PD powered, 4GiB NAND, 1GiB RAM, dual core arm64 iirc Apr 20 12:56:33 wow, gigabytes :) Apr 20 12:56:46 I want to build a HA dns/dhcp/.. setup with 2 (or more of those) Apr 20 12:56:52 i saw something on your blog Apr 20 12:56:56 but it was a few years old Apr 20 12:57:35 this would be standalone for DNS/DHCP, maybe radius for 802.1x/WPA-Enterprise Apr 20 12:57:44 still using those apu2 and erl for main router, for now Apr 20 12:58:06 i'm behind an apu2 here Apr 20 12:58:16 or at least, this irc box is Apr 20 12:58:54 ah that's a VM on a OVH box Apr 20 12:59:07 fortunately I was not in the cloud that went up in the clouds Apr 20 12:59:10 hehe Apr 20 13:02:14 the reason to build some things in tools is to avoid problems due to different versions of those tools in different distros afaik Apr 20 13:02:38 yes, i can imagine Apr 20 13:02:42 'and make sure you unset SED' :) Apr 20 13:04:05 with some luck this part of preparing the build system will be done just before i have to go away ;) Apr 20 13:05:35 it can take a while Apr 20 13:05:47 yep Apr 20 13:05:49 this is fine :) Apr 20 13:05:59 i suspect it's pretty clever about not rebuilding all the things all the time Apr 20 13:06:11 and i hope it'll benefit from my ccache Apr 20 13:07:01 it is, but sometimes you might hit annoying / undiscovered issues with that Apr 20 13:07:16 the buildbots start from a clean slate every time - but do use ccache iirc Apr 20 13:07:20 the powerdns code base has prepared me for make surprises ;) Apr 20 13:07:39 btw, how did you figure out i was here about powerdns? i don't have it in my cloak :) Apr 20 13:08:03 I'm using powerdns for a long time, the nick looked familiar Apr 20 13:09:13 that'll do it :) Apr 20 13:17:41 oh, it even builds a gcc for x86 Apr 20 13:25:00 ok, need to go, thanks so far stintel Apr 20 13:37:20 Habbie: ah, rebase worked, but now build fails on kea Apr 20 13:38:04 * stintel disables kea for now Apr 20 13:49:35 Habbie: FYI, o2h builds fine with the ruby dep removed Apr 20 14:10:48 stintel, thanks, i am not surprised :) Apr 20 14:11:39 i should come home later today to a bunch of built packages - or an error 5 seconds after i left Apr 20 14:12:15 haha, so true Apr 20 14:12:21 stintel, i wonder, is there even a way for a missing runtime dep to break the building of a package? Apr 20 14:12:59 yes Apr 20 14:13:09 where was it again Apr 20 14:13:49 Habbie: https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=include/package-ipkg.mk;h=c2017cd220622f88aa437b73f48b5d3ffc1535d6;hb=HEAD#l70 Apr 20 14:14:17 if something is linked to a lib of which the package is not in DEPENDS you'll hit that failure Apr 20 14:14:26 I just had that with kea Apr 20 14:18:03 right! it checks without running things Apr 20 14:18:30 debian has the opposite feature - generatijg runtime deps from an ELF read Apr 20 18:28:09 Did anyone avoided hacking their routers because of US DMCA (1201) provisions against circumventing any technological protection measures? Apr 20 19:26:03 Hauke: Don't care (EU resident here, thankfully) but… how is that crap even enforceable? Apr 20 19:26:34 (I mean, assuming no digital signatures and/or locked bootloaders.) Apr 20 19:51:53 stintel, came home to a bunch of built packages - dnsdist+deps in 35 minutes Apr 20 19:59:31 Habbie: nice! Apr 20 20:19:19 rsalvaterra: I am also located in the EU, but someone was asking me if I know someone affected by this, to help lobby against this. Apr 20 20:27:06 Hauke: The EFF would probably be interested in helping too, no? Apr 20 20:28:11 stintel, .. for the wrong architecture :D Apr 20 20:28:41 Habbie: oops. ath79? Apr 20 20:28:48 no, x86 instead of x86_64 Apr 20 20:28:53 it's not like i forgot Apr 20 20:29:10 but i did not look close enough at the subarch thing or whatever it's called Apr 20 20:31:22 oh well, i know how things work now (kind of) so in another 40-50 minutes i should be good (and also, asleep ;) ) Apr 20 22:08:28 rsalvaterra: the mt7621 interrupts are already manually tuned AFAIK Apr 20 22:09:00 mangix: They do look like it, yes. Apr 20 22:36:09 why is rcS still in inittab as '::sysinit:/etc/init.d/rcS S boot'? there is no /etc/init.d/rcS anymore Apr 20 22:36:52 zorun: Do you have some time to help with the 21.02.0-rc1 release notes? https://openwrt.org/releases/21.02/notes-21.02.0-rc1 Apr 20 22:36:55 or anyone else Apr 20 22:44:35 Hauke: these days, not so much, sorry :( and it's probably a fair amount of work Apr 20 23:13:31 zorun: ok, I will extedn it tomorrow Apr 20 23:20:20 Hauke: when is the release planned? I can proof-read it sometimes this week Apr 20 23:29:16 zorun: the build is currently running, I hope to get it out in ~24 hours or ~48 hours **** ENDING LOGGING AT Wed Apr 21 03:01:26 2021