**** BEGIN LOGGING AT Wed Mar 24 02:59:57 2021
Mar 24 03:21:48 <Grommish> this package as a Host/Install define.. how do I call just that from make rather than running Host/Compile?
Mar 24 03:47:34 <aparcar[m]> Grommish: make package/foobar/host/compile
Mar 24 03:49:05 <Grommish> aparcar[m]: Yeah, I don't want to call /compile, i want just the Host/Install section
Mar 24 03:49:34 <Grommish> i mean, I know Host/Compile will call Host/Install, but I didn't know if I can just call Host/Install
Mar 24 03:52:45 <Grommish> I've got issues with the script code in my Host/Install I'm trying to test and fix.  When I do a Host/Compile, it takes an hour to get to the Host/Install section :D
Mar 24 03:52:56 <Grommish> So you might understand why I was looking
Mar 24 04:15:04 <aparcar[m]> Grommish: true I do understand that
Mar 24 04:15:17 <aparcar[m]> No idea if that's possible, maybe you can add patches for that :)?
Mar 24 04:16:27 <Grommish> I think it used to be that way, but was changed to be how it is now.  Ah well, I'll work around it :)
Mar 24 04:24:58 <aparcar[m]> how long does rust take to compile? it seems a bit much when each run takes about 250 minutes
Mar 24 05:49:44 <Grommish> First run is long, but it creates the installation files, so that any future run will use the binaries already created
Mar 24 05:51:16 <Grommish> I'm working on the part that handles the install binary archives right at the end of the build.  But, I have to rebuilt the toolchain for it ot get to the point
Mar 24 05:51:59 <Grommish> If I could trigger just Host/Install without Host/Compile, i'd be ok, but *shrug* Ah well
Mar 24 06:08:05 <aparcar[m]> Grommish: mhh
Mar 24 06:08:24 <aparcar[m]> Grommish: maybe you can build the host system to add that feature
Mar 24 06:21:53 <rmilecki> Request for testing LuCI + DSA: https://forum.openwrt.org/t/request-for-testing-luci-on-dsa-devices/92126
Mar 24 13:27:12 <aep> someone remind mw how flash erase works. erase size 00010000 means the minimum writable block is actually 65536 bytes? that seems huge
Mar 24 13:27:52 <karlp> erase size is erase size, not "minimum writeable block" ?
Mar 24 13:28:32 <rsalvaterra> aep: For raw NAND, that's actually a small block size.
Mar 24 13:29:12 <aep> huh i had something like 4bytes in the back of my head
Mar 24 13:29:18 <aep> *4kb
Mar 24 13:29:27 <karlp> aep: from a random spi nor datasheet, pages are 256bytes, and erase size is 16 pages.
Mar 24 13:29:38 <aep> thanks
Mar 24 13:29:50 <aep> trying to split a partition smaller than that, but i guess that's a nope
Mar 24 13:30:03 <jow> afair OpenWrt has patches for that
Mar 24 13:30:15 <jow> to allow writes across part boundaries
Mar 24 13:30:17 <aep> oh it can do smaller partitions than erase size?
Mar 24 13:30:22 <aep> that sounds scary
Mar 24 13:30:53 <aep> i forgot how this works... you need to erase nand BEFORE writing right?
Mar 24 13:31:00 <jow> yeah, behind the scenes it'll read the entire block, mask out parts belonging to another part, set new contents, write back entire block
Mar 24 13:31:19 <aep> so if you loose power during that it's GG
Mar 24 13:31:21 <jow> and write back = erase, write
Mar 24 13:31:53 <jow> right. it is/was used for the kernel and rootfs partitions which are adjacent
Mar 24 13:32:28 <jow> and which are both non-vendor partitions
Mar 24 13:32:41 <jow> doing something like that next to a bootloader or caldata partition is rather... risky
Mar 24 13:32:59 <aep> i'm doing that to art. because i need a partition that's persistant D:
Mar 24 13:33:10 <aep> art has free space, but when you loose it it's rather not great
Mar 24 13:33:26 <aep> actually, stupid me, we never write this post factory. never mind
Mar 24 13:33:52 <aep> going to try setting a smaller size in the dts and see if the patch just magically works
Mar 24 13:42:24 * ldir  makes the mistake of wandering to the dnsmasq web site and sees there's a 2.85rc1 out.
Mar 24 13:42:54 <rsalvaterra> ldir: Bump it! :)
Mar 24 13:44:11 <ldir> hmmmmmmm
Mar 24 13:45:54 <ldir> dnsmasq releases mostly only come out for one reason.  All I'm saying.
Mar 24 13:46:53 <rsalvaterra> ldir: Hm… CVE?
Mar 24 14:07:47 <rsalvaterra> ldir: I just checked the dnsmasq git log, there are some really nice fixes in there.
Mar 24 14:13:15 <plntyk> aep new TLC/QLC NAND has up to 16-96MB erase block size according to https://www.anandtech.com/show/16491/flash-memory-at-isscc-2021
Mar 24 14:17:02 <rsalvaterra> plntyk: That's… all sorts of crazy.
Mar 24 14:22:16 <aep> jow: works magically out of the box, thanks
Mar 24 19:08:16 <philipp64> Hoping that PR #4006 on openwrt/openwrt will get merged before the end of the week...
Mar 24 19:08:24 <philipp64> It's a 3-line change.
Mar 24 19:11:46 <philipp64> aparcar[m]: ping
Mar 24 19:12:24 <philipp64> thanks for the work on PR #15195. had a question about that which I added as a comment just now...
Mar 24 19:12:50 <philipp64> also... been wondering what the [m] is about...
Mar 24 19:42:38 <aparcar[m]> philipp64: pong
Mar 24 19:42:46 <aparcar[m]> you're mentioning circleci, which isn't used anymore
Mar 24 19:45:00 <philipp64> I suspect the failures will likely re-emerge with whatever else we fire up to replace it.
Mar 24 19:45:29 <philipp64> gmpdh was demonstrably being built incorrectly...
Mar 24 19:45:50 <philipp64> our PIC support has been weak.
Mar 24 19:46:06 <philipp64> Hopefully this fixes it going forward.
Mar 24 20:01:29 <mangix> philipp64: PIC support can be checked with checksec.
Mar 24 20:03:08 <philipp64> no, I'm talking about build-time support.  certain source needs to be told it's being build in a position-independent way.
Mar 24 20:21:57 <ldir> philipp64: I could be persuaded to merge simply on the 'f**k it, let's see what blows up approach'
Mar 24 20:24:34 * ldir pulls in the PR, does make dirclean, starts a build from scratch
Mar 24 20:24:37 <aparcar[m]> ldir: please take over than
Mar 24 20:25:08 <ldir> there's always git revert
Mar 24 20:25:45 <philipp64> yup.
Mar 24 20:26:06 <philipp64> and the things that use it would blow up spectacularly if it was problematic.
Mar 24 20:26:20 <ldir> "3 line change" - what could possibly go wrong :-D
Mar 24 20:26:41 <philipp64> it would be like smoking in a fireworks factory... not like building a dam on a fault line...
Mar 24 20:27:27 <ldir> oh so you mean we get to enjoy the loud noises and pretty lights too - hell I'm convinced!
Mar 24 20:27:32 <philipp64> frankly I'm more surprised that this lack hasn't caused previous issues.
Mar 24 20:27:49 <philipp64> well, I mean we get to find out sooner rather than later.
Mar 24 20:28:11 <philipp64> a cigarette only lasts a couple of minutes at most.
Mar 24 20:28:17 <philipp64> a dam is there for decades.
Mar 24 20:28:21 <philipp64> okay, bad metaphor.
Mar 24 20:38:47 <mangix> let's see...
Mar 24 20:39:02 <philipp64> ldir: too bad that openwrt PR's don't cause builds to be triggered...
Mar 24 20:39:04 <mangix> checksec has a lot of dependencies
Mar 24 20:39:05 <mangix> nice
Mar 24 20:39:10 <philipp64> mangix: ?
Mar 24 20:39:17 <philipp64> ah.
Mar 24 20:39:23 <mangix> pkg_write_filelist: Failed to open //usr/lib/opkg/info/binutils.list: No space left on device.
Mar 24 20:39:27 <mangix> FUUUUUUUUU
Mar 24 20:40:15 <mangix> I should probably be building with GCC8 or something
Mar 24 20:42:58 <mangix> oh forgot GCC7 is still available
Mar 24 20:47:56 <Grommish> ldir: "I could be persuaded to merge simply on the 'f**k it, let's see what blows up approach" Having no idea what this refers to, I am strangely in favor of this approach for most things
Mar 24 20:48:27 <decke[m]> Grommish: good evening
Mar 24 20:48:42 <Grommish> decke[m]: *waves* Hey hey
Mar 24 20:50:07 <decke[m]> Grommish: just wanted to ask about the current status of your ER10X
Mar 24 20:52:17 <Grommish> decke[m]: Well.. There isn't a dsa driver for the rtl8367rb yet, although I have mixed swonfig and the dsa for the mt7621 board.  Ports 0-4 and 6-9 work.  The LEDs don't work forf 5-9, and POE probably doesn't work but I can't test because i don't have an injector at the moment
Mar 24 20:53:02 <Grommish> decke[m]: My goal is to look at the 8366_smi dsa driver and the 8367rb datasheet to see if it's similar enough to change the values in the 8366 file to make a DSA driver for 8367rb
Mar 24 20:54:07 <Grommish> and no one has bitten on whyh eth5 might not be working, although it seems its an swconfig issue and I"e just never used it before to know better.
Mar 24 20:54:17 <Grommish> I've got a post on the forum, but eh.. you can imagine
Mar 24 20:54:56 <decke[m]> Grommish: yeah I've seen that post and the silence
Mar 24 20:55:19 <decke[m]> sounds like the overall status is quite good
Mar 24 20:55:27 <Grommish> In the mean time, I can tell yhou the ER10x wil run suricata6 :)
Mar 24 20:55:43 <Grommish> but only because it's the current test bed I'm running hehe
Mar 24 20:56:11 <decke[m]> and I think I have seen some strange code in the ubiquiti sources for the LEDs
Mar 24 20:56:41 <Grommish> I softrware probed the gpio ports, they have a long lits of them
Mar 24 20:56:48 <Grommish> I want to do the same under EdgeOS and see what blink
Mar 24 20:59:13 <decke[m]> Grommish: you're really doing a great job :)
Mar 24 20:59:50 <Grommish> decke[m]: :) It's fun :)
Mar 24 21:02:04 <decke[m]> Grommish: about the POE support, those devices have passive poe out on the highest port and passive poe in on the first port
Mar 24 21:02:41 <Grommish> decke[m]: Right.. eth0 is PoE in and eth9 is PoE passthru
Mar 24 21:02:59 <decke[m]> Grommish: but you can also power the device with the wall plug and have passive poe out
Mar 24 21:03:07 <Grommish> but whether that is hardware or software I dunno yet
Mar 24 21:03:56 <decke[m]> Grommish: look at the EdgeRouter X wiki page, it describes the gpio setting that needs to be set
Mar 24 21:03:59 <owrt-snap-builds> Build [#24](https://buildbot.openwrt.org/master/images/#builders/15/builds/24) of `armvirt/64` failed.
Mar 24 21:06:28 <decke[m]> Grommish: https://openwrt.org/toh/ubiquiti/ubiquiti_edgerouter_x_er-x_ka#poe_out_on_edgerouter_x
Mar 24 21:08:23 <Grommish> Interesting, since PoE isn't on eth4 here, but eth0 and eth9.. I wonder how that will work
Mar 24 21:09:00 <Grommish> But I can use that probably ;0
Mar 24 21:24:16 <philipp64> Grommish: refers to PR #4006 on "openwrt".
Mar 24 22:08:09 <rsalvaterra> WTF, I made distclean, lost the packages repo configuration and now I don't remember how to recreate it! *facepalm*
Mar 24 22:08:59 <rsalvaterra> (Because I have the packages feed in a directory separate from the buildroot.)
Mar 24 22:09:51 <olmari> rsalvaterra: zfs rollback 😉 (sorry, had to 😜 )
Mar 24 22:10:20 <rsalvaterra> olmari: …
Mar 24 22:10:55 <rsalvaterra> I *think* it was symlinked somewhere…
Mar 24 22:14:08 <rsalvaterra> I think I got it.
Mar 24 22:17:39 <owrt-snap-builds> Build [#20](https://buildbot.openwrt.org/master/images/#builders/2/builds/20) of `layerscape/armv7` completed successfully.
Mar 24 22:17:41 <rsalvaterra> And now let's see how an i7-4770R compares to a Pentium D 950, building from scratch. B)
Mar 24 22:24:43 <owrt-snap-builds> Build [#17](https://buildbot.openwrt.org/master/images/#builders/60/builds/17) of `mpc85xx/p1010` failed.
Mar 24 22:47:49 <pkgadd> rsalvaterra: moving from Q9550 to i7-2600k was a night-and-day difference
Mar 24 22:48:58 <rsalvaterra> pkgadd: Well, this goes from about 3 hours to less than 30 minutes. Not bad.
Mar 24 23:05:38 <mangix> rsalvaterra: you should use ccache
Mar 24 23:07:30 <rsalvaterra> mangix: I thought about it… haven't used it since my Gentoo days…
Mar 24 23:08:15 <mangix> it makes a huge difference on slow systems
Mar 24 23:08:33 <mangix> actually even on fast ones...
Mar 24 23:09:37 <rsalvaterra> You pay with storage space, though… :)
Mar 24 23:12:18 <mangix> rsalvaterra: not that much
Mar 24 23:12:34 <mangix> du -sh .ccache/
Mar 24 23:12:34 <mangix> 196M    .ccache/
Mar 24 23:13:01 <mangix> it's so useful that meson defaults to it
Mar 24 23:13:05 <rsalvaterra> Compressed?
Mar 24 23:14:20 <mangix> ccache = compiler cache
Mar 24 23:15:42 <rsalvaterra> mangix: I know, I was asking if the cache is also compressed.
Mar 24 23:16:04 <rsalvaterra> (I could just RTFM, I know. :P)
Mar 24 23:16:43 <rsalvaterra> Right. https://ccache.dev/manual/4.2.html#_cache_compression
Mar 24 23:17:49 <mangix> oh cool
Mar 24 23:18:00 <mangix> didn't know that
Mar 24 23:18:39 <mangix> 5GB size limit by default. I'm sure you can spare that :P
Mar 24 23:19:04 <mangix> olmari: it's all about btrfs
Mar 24 23:19:16 <rsalvaterra> Yeah, 196 MB struck me as a bit small for OpenWrt builds… ;)
Mar 24 23:19:49 <urjaman> I had to adjust that down on my C201, ccache kept randomly eating all of my home :P (about third of the tiny eMMC is a bit much for a limit here lol)
Mar 24 23:20:31 <urjaman> but also, this thing isnt building openwrt, just some other things that use meson...
Mar 24 23:20:38 <mangix> olmari: funny thing about that actually. I was looking into NILFS2 and it doesn't offer functionality like that. You have to use rsync to replace the current working state with a snapshot...
Mar 24 23:21:23 <mangix> urjaman: setting CC/CXX explicitly disables ccache
Mar 24 23:22:00 <mangix> set -U -x CC clang for example disables ccache
Mar 24 23:23:06 <karlp> how do you only have 196M in ccache, mine's many gigs.
Mar 24 23:23:30 <karlp> what's your hit %?
Mar 24 23:24:58 <mangix> karlp: that was not from an openwrt machine
Mar 24 23:25:07 <karlp> ah :)
Mar 24 23:25:21 <mangix> du -sh .ccache/
Mar 24 23:25:22 <mangix> 4.3G    .ccache/
Mar 24 23:26:09 <mangix> That's probably shared between multiple platforms
Mar 24 23:39:06 <mangix> hmm fsffrance.org seems to be down
Mar 25 01:58:14 <OutBackDingo> whos up for some paid development work ? :)
**** ENDING LOGGING AT Thu Mar 25 02:59:56 2021