**** BEGIN LOGGING AT Wed Nov 18 03:02:03 2020
Nov 18 03:49:00 <FirstTime_NoTime> Any suggestions where to look? Porting my patches for a mt7620a board from 19.07.4 to master, I have lost access to the LED (now giving me error -22 when trying to access to gpio pin via /sys/class/gpio).
Nov 18 03:54:58 <FirstTime_NoTime> The reset button also doesn't work (same -22 error code when trying to assign the GPIO via DTS to gpio-keys-polled)
Nov 18 04:29:53 <aparcar[m]> mangix: do you handle the perl update?
Nov 18 05:05:10 <mangix> aparcar[m]: what perl update?
Nov 18 05:10:09 <aparcar[m]> i thought you suggested a perl update on the ml
Nov 18 05:11:13 <mangix> the person had a patch for it. I suggested that he post it on the packages repo.
Nov 18 05:23:03 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_tegra.html has been updated. (0% images and 100.0% packages reproducible in our current test framework.)
Nov 18 07:27:18 <ynezz> it's just me or the PKG_MIRROR_HASH stuff doesn't work properly for PKG_SOURCE_PROTO:=git? My test case is, that I remove the package foo tarball from the dl dir, make the hash intentionaly wrong, then do `make package/foo/{clean,compile}`, git clone is performed and build continues
Nov 18 07:29:00 <ynezz> I need to do `make package/foo/check` to get just the warning
Nov 18 07:29:26 <ynezz> why isn't hash mismatch hard error?
Nov 18 07:43:49 <ynezz> it works almost as expected for PKG_HASH, when the tarball is missing, it will download and check
Nov 18 07:44:35 <ynezz> if I change the tarball content on the disk (or that hash) then it happily continues building
Nov 18 07:45:00 <ynezz> shouldn't we always perform that check step and make that warning hard error?
Nov 18 07:46:12 <ynezz> I for example would prefer sane sources in exchange for little more computing power spent on checksums
Nov 18 07:57:21 <Grommish> if it hard-errored, it would never checkout the new source, would it?
Nov 18 07:57:48 <Grommish> ynezz: I'm having issues with download and unpack for non-package files
Nov 18 08:32:53 <rsalvaterra> 'morning!
Nov 18 09:03:32 <nitroshift> rsalvaterra, o/
Nov 18 09:04:07 <mangix> ynezz: good question
Nov 18 09:04:14 <rsalvaterra> nitroshift: o/
Nov 18 09:12:50 <rsalvaterra> aparcar[m]: Patch sent. :)
Nov 18 09:33:28 <Tusker> heya guys, I just rebuilt my build box (moved from ceph to zfs), but libubox is failing to build, is it just me, or is it expected ?
Nov 18 09:36:35 <Tusker> hmm... google says that it's a build directory issue... make distclean time
Nov 18 10:20:30 <Tusker> zfs with cache on ssd is much faster than ceph with tiered ssd... glad I took the time to rebuild my server
Nov 18 10:31:14 <Tusker> time make - real2m37.375s for a no-changes rebuild, vs >5minutes on ceph
Nov 18 14:01:42 <barhom_> I have a firewall.user line that looks like this, "iptables -t mangle -I PREROUTING -i br-lan -m physdev --physdev-in wlan0-1 -j MARK --set-mark 1". I would like to rewrite it into a /etc/config/firewall config instead. But I'm not sure how exactly, especially with the physdev requirement
Nov 18 15:37:44 <owrt-1806-builds> build #90 of rb532/generic is complete: Failure [failed updatefeeds]  Build details are at http://buildbot.openwrt.org/openwrt-18.06/images/builders/rb532%2Fgeneric/builds/90
Nov 18 15:56:33 <jow> barhom: you could use `option extra '-m physdev --physdev-in wlan0-1`
Nov 18 15:57:36 <jow> barhom: https://pastebin.com/iawpTJB9
Nov 18 15:59:21 <jow> ynezz: iirc PKG_MIRROR_CHECKSUM was always meant to soft fail
Nov 18 15:59:58 <jow> ynezz: it originally was designed as performance improvement (ability to fetch a precompressed cached tarball) to avoid potentially slower git cloning
Nov 18 16:00:24 <jow> it does not provide the smae strong integrity guarantees as the normal checksum for tarball downloads
Nov 18 16:03:26 <jow> the idea is that if the cached tarball is either absent or mismatching, a clone operation is performed using the specified git commit hash, which guarantees source integrity
Nov 18 16:16:40 <Guest76643> exit
Nov 18 16:22:57 <owrt-1806-builds> build #91 of rb532/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/openwrt-18.06/images/builders/rb532%2Fgeneric/builds/91
Nov 18 16:41:35 <ynezz> jow: so the mismatch part is currently not working: rm dl/libubox-2020-08-06-9e52171d.tar.xz; make package/libubox/download V=sc; sed -i 's/PKG_MIRROR_HASH:=c/PKG_MIRROR_HASH:=d/' package/libs/libubox/Makefile; make package/libubox/{clean,compile} V=sc
Nov 18 16:42:02 <ynezz> jow: it wouldn't download the package again even if it has now the wrong hash
Nov 18 16:42:21 <ynezz> jow: is my observation correct?
Nov 18 16:46:08 <ynezz> well, by re-downloading I mean git cloning again
Nov 18 16:53:21 <jow> you mena when changing the hash after the cached tarball has been downloaded?
Nov 18 16:53:45 <jow> if so then yes, that should be fixed
Nov 18 16:59:26 <ynezz> yes, will take a look at it
Nov 18 17:35:47 <damex> any suggestion on some openwrt supported 'l3 switch' that is not limited to 1gbit to soc? edgerouter X have 1gbit.  edgerouter X-sfp and mikrotik hex-s have sfp port on a separate 1gbit to soc
Nov 18 17:35:59 <rr123> debugging a autogen.sh issue for hours, 'staging_dir/host/bin/aclocal: 2: aclocal.real: not found' ..aclocal does refer to aclocal.real which is nowhere to find
Nov 18 17:36:51 <rr123> https://nopaste.xyz/?4632386ba2223b5a#29GHeMGHWsJvb1P59tPKNRRNsoZic93YzASjDC1AeDY9
Nov 18 17:36:52 <damex> i picked hex-s to see how well it performs (already have er-x) - need another recommendation for a well supported platform :)
Nov 18 17:39:49 <rr123> builderr log: https://nopaste.xyz/?cecebcc4d76ad5bc#4DTLrAMtZYu4aeS8UyEkVD9X4Nu7YLPmNALuuvGGfYaG
Nov 18 17:41:54 <rr123> repost the build error log: https://nopaste.xyz/?4b5cf07772fc147b#79qoxx57UsxYXBqz3ifvbuFPg16N4rPHnFXVXChUDJTo
Nov 18 18:11:21 <aparcar[m]> rsalvaterra: thanks
Nov 18 18:38:26 <aparcar[m]> jow: if you want you can criticize some poorly written shell scripts here https://github.com/openwrt/openwrt/pull/3599
Nov 18 18:48:28 <damex> aparcar[m]: did you already try shellcheck? also maybe you can try readon file to the memory <once> and then try to parse it?
Nov 18 18:48:47 <jow> aparcar[m]: I like it. Only minor suggestion would be to support wget as fallback
Nov 18 18:49:05 <jow> but thats no blocker for me, just to align with download.pl which supports either curl or wget
Nov 18 18:50:01 <aparcar[m]> damex: who do I load it into memory? Just inside a variable? The performance is somewhat terrible, that's why packages.git isn't enabled by default
Nov 18 18:50:09 <aparcar[m]> jow: I'll look into that thanks
Nov 18 18:50:27 <damex> aparcar[m]: yeah,  inside variable.
Nov 18 18:50:33 <jow> you should also be able to pass data.tar.gz as addiitonal argument to tar tzf directly
Nov 18 18:50:36 <jow> no need for grep
Nov 18 18:51:17 <jow> and some sed command to parse the feed list should be way faster than the read / case loop
Nov 18 18:52:08 <jow> but likely neglectible, the script is suppoed to be run on the host after all, and that is doing way more complex things during compilation
Nov 18 19:33:04 <rsalvaterra> aparcar[m]: You're welcome!
Nov 18 19:51:31 <aparcar[m]> jow: i'm curious about the sed command, you'd basically use it with some multi line mode?
Nov 18 20:01:08 <aparcar[m]> damex: I stored it in a variable but it didn't speed up anything. In a later iteration I'm happy to use sed
Nov 18 20:01:21 <aparcar[m]> however the multiline sed stuff looks like it's own language
Nov 18 20:29:20 <aparcar[m]> why is openvpn in openwrt.git?
Nov 18 20:35:12 <rsalvaterra> aparcar[m]: Because it's considered a core package…?
Nov 18 20:35:46 <aparcar[m]> it's not really necessary for the default installation, that's why i'm wondering
Nov 18 20:39:37 <Guest88411> greetings, i have a few questions about MT76 driver
Nov 18 20:40:28 <Guest88411> specifically regarding its stability since OpenWrt 19.07
Nov 18 20:40:52 <Guest88411> has it gotten better? because the one included with 19.07 is unstable and slow.
Nov 18 20:41:49 <aparcar[m]> yes now it works thanks
Nov 18 20:42:00 <aparcar[m]> wrong channel
Nov 18 20:42:02 <Guest88411> this is for the MT7621 platform and specifically the mt7615n wifi chip
Nov 18 20:42:38 <Guest88411> this is the openwrt-devel channel right ?
Nov 18 20:42:49 <PaulFertser> Guest88411: it's being constantly improved.
Nov 18 20:43:04 <PaulFertser> Guest88411: and so yes, it should be better now.
Nov 18 20:43:27 <Guest88411> is there an eta when the pulls will be included (like a release date for 20.whatever)
Nov 18 20:43:54 <Guest88411> and would the new improvements be included in recent snapshots?
Nov 18 20:47:21 <PaulFertser> Guest88411: improvements are included in snapshots.
Nov 18 20:56:37 <Guest88411> PaulFertser Thanks.
Nov 18 20:57:36 <PaulFertser> Guest88411: if you're interested in specifics, you can just see the git history, package/kernel/mt76/Makefile
Nov 18 21:00:31 <aparcar[m]> how can I set compile definitions (-D) for non cmake packages?
Nov 18 21:03:25 <aparcar[m]> EXTRA_CFLAGS+=-D_GNU_SOURCE apparently
Nov 18 21:08:45 <aparcar[m]> mangix: can you help me out here? https://github.com/openwrt/packages/pull/13944
Nov 18 21:14:30 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_x86.html has been updated. (100.0% images and 100.0% packages reproducible in our current test framework.)
Nov 19 02:20:04 <mangix> aparcar[m]: why you need GNU_SOURCE?
Nov 19 02:20:29 <aparcar[m]> mangix: no was just an example. I needed the getopt full
Nov 19 02:29:18 <mangix> aparcar[m]: enabling GNU_SOURCE glogally is not a good idea.
Nov 19 02:29:28 <mangix> some packages get an increase in size
Nov 19 02:29:44 <aparcar[m]> mangix: that's just an example I copied from a differen package
Nov 19 02:29:59 <aparcar[m]> I replaced it with HAVE_GETOPT_LONG (as seen in the package)
Nov 19 02:34:42 <mangix> armbian uses felix' mvebu patch
Nov 19 02:34:44 <mangix> why...
Nov 19 02:35:04 <mangix> they even have it under a needs-checking folder
Nov 19 02:35:05 <mangix> rofl
Nov 19 02:36:44 <aparcar[m]> mangix: could you test the imagebuilder signature check for me?
Nov 19 02:37:01 <aparcar[m]> it's a release blocker for 20.x so I'd like to have some external testing
Nov 19 02:39:30 <mangix> link?
Nov 19 02:41:51 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_ar71xx.html has been updated. (99.1% images and 100.0% packages reproducible in our current test framework.)
Nov 19 02:41:59 <aparcar[m]> mangix: https://github.com/aparcar/openwrt/commits/staging
**** ENDING LOGGING AT Thu Nov 19 02:59:59 2020