**** BEGIN LOGGING AT Mon Oct 08 03:00:00 2018 Oct 08 05:08:06 morning *ywan* Oct 08 05:08:10 *yawn Oct 08 07:00:15 Hello All! Oct 08 07:00:48 Since 18.06.1 is there any very useful change for Archer C7 v2 in the trunk? Oct 08 07:23:19 xcom169: i have on on lede r707 that seems to be working fine ;-) Oct 08 07:23:42 uptime 87 days Oct 08 07:36:54 hi Oct 08 07:45:07 blogic: o/ Oct 08 07:45:24 blogic: couldn't reproduce the issue with RPS balancer on my ERL, might be x86 only Oct 08 07:46:44 my guess is that the code is racy with multi-queue rx Oct 08 07:46:53 the timer setup looks fragile Oct 08 07:47:12 i'm currently thinking about alternative approaches that might also be cheaper in terms of cpu cycles Oct 08 07:48:38 russell--: jow: what's the state of ubifs corruptions? Oct 08 08:02:23 stintel: just revert it please Oct 08 08:02:27 i dont mind Oct 08 08:02:41 blogic: ok Oct 08 08:03:48 done Oct 08 08:11:56 rmilecki: still there, i just looked again. i'm kind of stalled waiting for a buildable 4.19-x kernel. Oct 08 08:12:47 russell--: so it's a regression introduced after 4.9, present in 4.14, not sure if present in 4.19, right? Oct 08 08:12:55 yes Oct 08 08:13:35 russell--: and you boot, edit some file, wait a minute or so, reboot and then that file is corrupted or something? Oct 08 08:13:46 s/reboot/power cut/ Oct 08 08:14:20 yeah, basically. for me only files that aren't changed from /rom/etc/config/ don't make it to flash Oct 08 08:14:37 the one that got my attention was /etc/config/dropbear Oct 08 08:14:50 since it kills the ssh server Oct 08 08:14:54 and i can't get in Oct 08 08:15:03 ok, thanks Oct 08 08:15:30 there is a cp -a somewhere in the first boot code that does that copy Oct 08 08:16:15 i can echo "foo" > somefile.txt and that seems to survive a power cut if i give it a few seconds Oct 08 08:17:04 so it may be overlay related... Oct 08 08:17:47 yes Oct 08 08:19:11 i'm not sure why that copy even happens, frankly. it wouldn't be necessary since the files are unchanged from the underlying /rom/ versions Oct 08 08:20:00 russell--: I think I recall one "cp -a" in fstools for switching from ram overlay to jffs2 overlay Oct 08 08:20:03 not sure if there are more Oct 08 08:20:21 yeah, but i think in the ubifs case, there isn't a ram overlay stage Oct 08 08:20:47 right Oct 08 08:22:31 the other thing to note is a manual "sync" from a shell cures the problem Oct 08 08:30:49 hmm, just tried this, yanking power at 90: for i in $(seq 1 100) ; do echo $i ; echo $i > /root/$i ; sleep 1 ; done Oct 08 08:31:56 the last file with size > 0 is 56, the last one with actual data (i.e. non-null) is 46. 57 onwards has a filename but length zero. Oct 08 08:32:36 last file is 87 Oct 08 08:33:13 but in the case of the /etc/config/dropbear and friends, waiting indefinitely doesn't help Oct 08 08:33:22 up to a few months Oct 08 08:34:56 (testing on a ubnt erx) Oct 08 08:53:42 'lo Oct 08 09:11:04 how are .ko module dependencies usually specified? Is it done by kbuild magic which looks up the unresolved symbols or is there a more explicit way? Oct 08 09:12:01 does anyone have working kernel 4.16-4.19 support that can be recommended? Oct 08 09:12:09 before I start Googling and find some semi-working one Oct 08 09:12:23 i've to test my spi fix on more recent kenel Oct 08 09:16:05 found this https://github.com/openwrt/openwrt/pull/1386 Oct 08 09:20:07 i honesty hate working with GitHub.... every single time I have to find anything there Oct 08 09:20:21 so that pull request is made for sartura:initial_kernel_4.19_support Oct 08 09:20:37 and there is no URL visible for the sartura repository Oct 08 09:20:46 how the heck am I supposed to check out that? Oct 08 09:21:14 https://github.com/sartura/ find the repository manually :/ Oct 08 09:21:27 and then you can use initial_kernel_4.19_support from that repo Oct 08 09:21:35 maybe there's a better way Oct 08 09:21:57 and good to know, I could use that for mesongx target :) Oct 08 09:22:48 ok, got it, thanks Oct 08 09:22:53 git clone -b initial_kernel_4.19_support git@github.com:sartura/openwrt.git Oct 08 09:25:43 rmilecki: you can fetch any PR as pull/$PRID/head, e.g. git fetch https://github.com/openwrt/openwrt.git pull/1386/head:PR1386 Oct 08 09:25:53 rmilecki: then you can work with "PR1386" as a branch Oct 08 09:26:22 jow: thanks! Oct 08 09:26:24 rmilecki: see also https://github.com/KanjiMonster/maintainer-tools/blob/master/github-apply.sh Oct 08 09:26:44 which you can use like ./github-apply.sh 1386 Oct 08 09:27:18 the token part is optional, you only need it if you want to auto-close prs after merging Oct 08 09:28:02 it's still a big-unknown for me, why they won't work on some friendly UI Oct 08 09:28:38 also at https://github.com/openwrt/openwrt/pull/1386 - scroll down to the end and click "command line" in "Use the web editor or the to resolve conflicts." Oct 08 09:28:49 that'll give you the cli commands for fetching the pr branch Oct 08 09:29:31 heh just tried to get that kimocoder rtl8814au driver going on a arm64 rock64, I guess the kernel_neon_begin and end points don't cover everything Oct 08 09:29:41 or alternatively the "command line instructions" in " Oct 08 09:30:13 "or view command line instructions. " right next to the "Merge pull request" button Oct 08 09:31:30 jow: hm, I don't see that "command line" Oct 08 09:32:34 am, I missing write access Oct 08 09:32:53 ah I see Oct 08 09:32:55 "Only those with write access to this repository can merge pull requests." probably that's th reason Oct 08 09:32:58 nevermind Oct 08 09:33:08 i got the URL, that's enough for me for now :) Oct 08 09:33:19 or merge this tree https://github.com/sartura/openwrt/tree/initial_kernel_4.19_support Oct 08 09:33:40 right, either make a shell alias like pr() { git fetch https://github.com/openwrt/openwrt.git pull/$1/head:PR$1 } or use the script I linked for the future Oct 08 09:33:51 will make working with prs easier Oct 08 09:43:12 I've repeatdly asked github to enable those command line display promptos for non-repo owners, Oct 08 09:43:19 they've repeatedly ignored it Oct 08 09:45:30 ok, so that Merge Request contains non-working code, because it references linux-4.19.0.tar.xz which doesn't exist yet Oct 08 09:45:39 so i tried switching to the: Oct 08 09:45:41 LINUX_VERSION-4.19 = -rc7 Oct 08 09:46:09 but that makes OpenWrt try to get linux/kernel/v4.x/testing/linux-4.19-rc7.tar.xz Oct 08 09:46:37 a correct URL is https://git.kernel.org/torvalds/t/linux-4.19-rc7.tar.gz Oct 08 09:46:46 what i did was download the tar.gz from http://kernel.org Oct 08 09:46:54 then i converted it to tar.xz Oct 08 09:47:05 sha256sum ed it Oct 08 09:47:21 etc Oct 08 09:47:40 and then the builder does go to download it it just uses the one in the dl folder Oct 08 09:47:40 i think OpenWrt buld system knows that rcX requite different URL Oct 08 09:47:45 I guess it could be teached to also try .tar.bz2 Oct 08 09:47:50 but I'll try that workaround for now Oct 08 09:48:55 but yeah, i tried to do this for an allwinner h6 device, i got far enough that i got an image to flash to an sd card but it still wouldn't boot Oct 08 09:55:23 Can anyone see anything screamingly wrong with https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=cad8330e5116e02ff0fcc188f8c1f05650cbb635 before I think about committing it. I usually get 'don't implement it that way' responses ;-) Oct 08 09:56:48 oh and I have a samsung 'smart' TV that wants to claim 'localhost' for its own. smart my backside! Oct 08 10:04:27 haha Oct 08 10:07:14 oh, great, another thing with that Pull Request for 4.19 kernel: Oct 08 10:07:15 1 out of 20 hunks FAILED -- saving rejects to file net/ipv6/ip6_tunnel.c.rej Oct 08 10:07:22 target/linux/generic/pending-4.19/666-Add-support-for-MAP-E-FMRs-mesh-mode.patch using plaintext Oct 08 10:08:36 hah, of course Oct 08 10:08:43 it's a PITA every time ;) Oct 08 10:09:05 must be the 666 Oct 08 10:09:19 ^^^ haha I just saw that too! Oct 08 10:10:02 beware 664 & 668 as they're neighbours of the beast Oct 08 10:16:42 oooh, just noticed something odd... is it just me? luci network->dhcp & dns shows the active & static leases on every tab (general - resolv & hosts files - tftp settings(!) - advanced settings. Oct 08 10:18:38 probably should only appear on the resolv & hosts files tab. Oct 08 10:22:26 ldir:dnsmasq patch looks fine to me; only remark is that you've --dhcp-ignore-names entry by default even if dhcp_name_ignore is empty Oct 08 10:24:17 dedeckeh: I checked that dnsmasq doesn't barf in that circumstance and it doesn't. Agree dnsmasq.init should probably only put in the dhcp-ignore-names if the list exists.... I'll have to go through other code to see what I can copy & paste to do that. Oct 08 10:24:46 * ldir is currently shuddering at dhcp.lua Oct 08 10:24:46 ldir: mostly that people taking over an older config will miss those, and nobody will bother to add them - maybe have a ignore common names option that defaults to 1 that adds wpad and localhost? maybe always add localhost? Oct 08 10:24:54 ldir:you could add [ -n "$dhcp_name_ignore" ] && xappend .. but it's a minor detail Oct 08 10:25:08 oh that easy! Oct 08 10:27:48 https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=201720180SB327 Oct 08 10:27:53 KanjiMonster: that's a variation on the debate that occurred on the dnsmasq list... should dnsmasq bake in those names. Oct 08 10:28:07 California introduced a bill to forbid devices with default password ^ Oct 08 10:28:38 non-unique default password Oct 08 10:28:39 stintel: is no password a default? Oct 08 10:28:49 good question :P Oct 08 10:28:56 ldir: no password and forcing the user to set one is allowed Oct 08 10:29:03 but we don't offer devices, so don't think it's relevant anyway Oct 08 10:29:16 but I think it's actually a good thing Oct 08 10:29:28 yeah Oct 08 10:29:39 but does this also apply to backdoor logins? Oct 08 10:29:44 :D Oct 08 11:02:14 ldir: active leases appearing on every tab is "normal" Oct 08 11:02:25 will eventually fix it if I get around to it Oct 08 11:05:33 jow: ok, not just me :-) Oct 08 11:20:48 i'll appreciate reviews of [PATCH] kernel: fix downloading rcX releases Oct 08 11:20:57 i'm far from being Makefile expert Oct 08 11:23:40 jow: is there somethign like section.addremove that enforces only a single section? Oct 08 11:23:58 I know that uci in gneeral doesn't really enforce that, but is there a helper in cbi for it? Oct 08 11:34:39 you mean zero or one? Oct 08 11:36:30 rmilecki: is there no way to get .tar.xz files for rcX releases? Oct 08 11:37:11 nbd: not that I'm aware of Oct 08 11:37:16 https://www.kernel.org/ links to .tar.gz only Oct 08 11:37:34 https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git lists .tar.gz files only Oct 08 11:37:55 I tried manually modifying URL to https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/snapshot/linux-4.19-rc7.tar.bz2 but it doesn't work Oct 08 11:39:52 ok Oct 08 11:40:34 jow: yeah, zero or one, but not multiple Oct 08 11:42:59 karlp: easiest solution is a named section with .addremove = true Oct 08 11:43:13 that requies picking a name though Oct 08 11:44:50 yay, RaZberry2 Z-Wave+ module on my ODROID C2 running OpenWrt is working \o/ Oct 08 11:45:39 jow: would actually work jsut fine for this case. I'll close the PR I just opened and retest it like that, no big deal moving from typed section to named section for that at least. Oct 08 11:46:10 or, nvm. you just merged it already :) Oct 08 11:50:02 hrm, moving from anon to named would need a ucidefaults to rename the section too. Oct 08 11:50:11 maybe just leave it as "user don't do that please" Oct 08 11:54:47 rmilecki: https://github.com/buildroot/buildroot/commit/ff4cccbdcf5417a5a8c251ff302791a306023cc1 Oct 08 11:54:47 linux: Fix URL for release candidate versions Oct 08 11:55:22 Might be relevant Oct 08 11:58:12 luaraneda: yeah, looks similar Oct 08 12:14:10 KanjiMonster: and everyone else who expressed an opinion https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=b21a88de1e51c678a82f4fc5484cd1a5842e51de better? Oct 08 12:47:40 ldir:looks good Oct 08 12:50:28 join #ansible Oct 08 12:50:36 although the 'or' bit of [ -n "$dhcp_name_ignore" -o "$dhcp_ignore_common_names" -gt 0 ] isn't working correctly. Oct 08 12:52:45 heh, I got cought on that recently. Oct 08 12:53:02 -o and -a don't short circuit the way you'd eexpect, they evaluate all first. Oct 08 12:53:24 so I did [ -n "$blah"] || [ "$blah" == "whatever" ]] Oct 08 12:55:34 hmm, I'll give that a go :-) it concerns me that '-o' appears in dnsmasq.init all over the place. Oct 08 13:12:31 nope - it turns out "$dhcp_name_ignore" is blank, yet the preceding config_list_foreach "$cfg" dhcp_name_ignore append_dhcp_ignore works ok Oct 08 13:16:29 aha! Oct 08 13:22:31 right a final attempt, this one at least works as intended. https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=c2d664f67383faf18f084120bc4b25f9560024b7 Oct 08 13:25:57 Hello Oct 08 13:26:13 I'm trying to support a Strong router (Mediatek MT7621, Dual band 1200, USB 3.0, microSD, ...) but I always receive a Kernel crash when booting (INFO: rcu_sched self-detected stall on CPU) Oct 08 13:26:22 https://pastebin.com/YVxagEVY Oct 08 13:26:35 Any idea why is happening? Oct 08 13:36:29 dedeckeh: ping Oct 08 13:40:50 thinking about this a & what KanjiMonster suggested...I've come to the conclusion that asking for localhost is inherently evil and handing out a lease associated with that name just shouldn't happen. the hostname wpad should by default be denied but optionally allowed. and adding other exclusions is down to config options. This actually simplifies the logic! Oct 08 13:42:24 asking for localhost should trigger iptables rules that isolate the device / disable the switch port ;) Oct 08 13:42:56 and electrocute the owner :-) Oct 08 13:43:37 There will be yet another commit once I've worked out the commit wording....the hardest bit! Oct 08 13:45:11 true. and making sure there are no spelling errors. and wether you did check that everything works as expected. and you didn't overlook some trivial case etc Oct 08 13:48:51 i think asking for localhost should be treated the same as asking for an address with no name Oct 08 13:49:50 I would assume that this is exactly what the option does Oct 08 13:49:56 you get a lease...you just don't get your preferred hostname...put into DNS. Oct 08 13:50:09 ldir:pong Oct 08 13:50:19 above ^^^^ Oct 08 13:52:14 ldir:you can count me in your conclusion Oct 08 13:55:44 https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=a6c51dcb8f9eaee92a5c9d8066520fd1dfe47c02 Oct 08 13:58:40 meh, missing netifd and uci already Oct 08 14:09:43 ldir: to add some other opinion to the mix; imho having the list of "default names considered evil" in uci is overkill. Personally I would put the "dhcp-name-match" statemnts into a native dnsmasq config snipped and assign them a more explicit tag name, like "bogus_dhcp_hostname" Oct 08 14:10:12 ldir: then only expose a high level default-on bool option which merely inserts "dhcp-ignore-names=tag:bogus_dhcp_hostname" Oct 08 14:10:38 which would work similar to the unnescessary winnt name filter Oct 08 14:11:13 (filterwin2k) Oct 08 14:11:28 ok umm, who has some time to help me with header nonsense? (typical redefinitions in in.h et al) Oct 08 14:11:44 jwh: usually it is enough to swap the inclusion order Oct 08 14:12:01 make sure that kernel headers are included first (or was it last? forgot already...) Oct 08 14:12:14 I an't figure it out, it's like a circular Oct 08 14:12:16 include Oct 08 14:12:39 it builds on the host fine, but when building as an openwrt package with glibc it just goes wrong Oct 08 14:13:04 gave up on making it work with musl as I don't know enough about the APIs to replace the ucontext stuff Oct 08 14:13:42 jow: I think you mean a bit like rfc6761.conf Oct 08 14:13:47 ldir: exactly Oct 08 14:14:03 maybe we could even reuse that file and rename it to defaults.conf or sio Oct 08 14:14:09 *so Oct 08 14:14:32 but just my opinion, I'm fine either way Oct 08 14:15:11 what is the 'pressure' of not having stuff in uci? Oct 08 14:15:37 my thinking was: Oct 08 14:16:00 - updated uci defaults are easily missed, especially for commonly customized configs Oct 08 14:16:26 - having more or less static value collections in uci feels suboptimal to me Oct 08 14:17:19 jow: also uh, what do you make of https://bugs.openwrt.org/index.php?do=details&task_id=1749 Oct 08 14:17:25 - and its alternative of having hardcoded defaults which can be (partly) reverted in uci feels suboptimal as well Oct 08 14:17:57 it's still present in master, the fix in the report works Oct 08 14:18:07 jwh: its a meh, glibc thing Oct 08 14:18:14 yeah Oct 08 14:18:19 I asked with https://github.com/openwrt/luci/pull/2207 in mind as well you see :-) Oct 08 14:18:51 I'd rather not go anywhere near glibc but heh Oct 08 14:19:28 crappy (supposedly, due to glibc specific things) but fast code, or old code that still builds on musl Oct 08 14:21:10 ldir: one minor nitpick: you could call append_dhcp_ignore as well for localhost and wpad instead of duplicating the code from there Oct 08 14:28:04 KanjiMonster: sure - but I may be just about to throw it all out anyway :-) Oct 08 14:28:43 ldir: why that? seemed like a reasonable change Oct 08 14:29:09 jow: so whats the official stance? glibc there but pretty much unsupported now? Oct 08 14:29:46 jwh: wouldn't say that - but among all the open tickets, its at the far low end of at least my todo list Oct 08 14:29:55 oh, right Oct 08 14:30:01 KanjiMonster: it's jow suggestion of putting the unwanted names in a file, instead of uci config. Oct 08 14:30:52 ldir: do you know offhand how dnsmasq reacts to undefined tags? Oct 08 14:31:13 e.g. dhcp-ignore-names=tag:doesnotexist - will it bail or silently ignore it? Oct 08 14:31:47 it silently ignores - I tested it. Oct 08 14:32:34 or more accurately, the tag doesn't evaluate true 'cos there's nothing to set it. Oct 08 14:33:02 who can I throw some money at? Oct 08 14:33:06 :D Oct 08 14:33:25 jwh: do I have to *do* anything? Oct 08 14:33:30 dance Oct 08 14:33:34 oh, I mean, code Oct 08 14:33:58 dammit, that was me nearly in there! Oct 08 14:34:03 lol Oct 08 14:35:18 nearly got my finest Theresa May impression! Oct 08 14:35:22 hha Oct 08 14:35:24 haha Oct 08 14:35:37 honestly, I would *really* like to get this done, i've been messing with it since july Oct 08 14:35:40 getting old now Oct 08 14:35:49 so, money talks right? Oct 08 14:36:35 jwh:I've a fix for that issue; it's quite trivial Oct 08 14:36:51 uh huh? Oct 08 14:37:16 jwh:the ef2progs issue Oct 08 14:37:19 oh Oct 08 14:37:30 I fixed that Oct 08 14:37:55 back to accel-ppp horribleness now... (but would be really nice to get into openwrt properly) Oct 08 14:44:49 KanjiMonster: https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=55f22b23b7a34bc4a8482260f1c1976237b1e591 until I change it completely...again :-) Oct 08 14:46:57 * ldir takes der Hund for a walk. Oct 08 14:50:00 ldir: that's "den Hund", you need to use the accusative case ;p Oct 08 15:18:53 alright so Oct 08 15:19:02 theres no conflicting headers Oct 08 15:19:27 no typical case of libc and kernel headers being included, yet they are Oct 08 15:19:38 without an include tree theres no way to figure out whats being pulled in where Oct 08 15:20:49 so uh, can I convince a single package to be built with the -M flag or something? Oct 08 15:21:36 but I still don't understand why the conflicts only appear within the openwrt build system Oct 08 15:21:46 and not on any other generic glibc system Oct 08 15:38:01 http://ix.io/1oGh Oct 08 15:38:14 even #ifdef'ing the defines doesn't prevent it being pulled in Oct 08 15:47:29 deffo give up now, ubuntu it is Oct 08 15:50:00 if there was actually a genuine problem I'd probably put more effort into it, but a couple of months chasing what looks like a build system problem since it doesn't appear *anywhere* else, enough is enough Oct 08 17:07:07 Anyone with MT7621 experience? I get a Kernel Panic when trying to port the Strong 1200 router Oct 08 17:19:00 _lore_: ^^ Oct 08 19:59:31 thoughts please https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=fb4c54635d8f90feb0fd236195e7f2c6781c25b9 vs https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commitdiff;h=237871a9da59eba245c6bb3f601fc6cf9438c72c Oct 08 20:01:07 the first one, just a list in /etc/config/dhcp right? Oct 08 20:01:27 effectively yes Oct 08 20:01:30 hm Oct 08 20:01:51 ok so the obvious one is.... how do you negate the "common" hostnames? Oct 08 20:02:21 if you can't, maybe defining them like the second one is better Oct 08 20:02:48 you can't disable 'localhost' by design, but you can the others with dhcp_ignore_common_names 0 Oct 08 20:03:03 hmmm Oct 08 20:03:13 personally I'd prefer the second one Oct 08 20:04:48 why? Oct 08 20:05:32 if more hostnames are added to the dhcp_ignore_common_names in future it's not easy to negate them Oct 08 20:05:35 people do weird stuff Oct 08 20:07:20 you disable the common_names option and add your own preferred list to option list dhcp_name_ignore Oct 08 20:07:39 oh hm, I guess Oct 08 20:07:55 first one then Oct 08 20:08:49 interesting. jow set me on the path to the 2nd version, which is smaller. Oct 08 20:09:30 and very similar to the RFC67?? implementation (which I also did) Oct 08 20:09:50 well honestly the idea of seperating out config portions is IMO nicer, but this particular thing seems pretty simple Oct 08 20:10:54 I hadn't considered the space difference, is it significant? Oct 08 20:13:21 no, not at all. the 2nd has less code for me to screw up :-) Oct 08 20:13:29 haha Oct 08 20:13:50 fair point, 2nd it is :D Oct 08 20:17:33 I think what bugs me about the 2nd is that it's directly including a config file...with all the potential syntax screwups that people could put in there. Oct 08 20:18:12 or indeed any other option. Same as the the RFC67?? file. Hmmm. Oct 08 20:18:25 oh dear I've got a stutter Oct 08 20:20:08 and here's a fun thing - all the xappend "--optionblahblah" lines.... well xappend strips off the leading '--' so why the bloody hell are they there in the first place ? Oct 08 20:20:18 lol Oct 08 20:20:33 maybe just generically specify the err, conf-dir option I think it is? Oct 08 20:20:38 and throw things in there to be included Oct 08 20:28:45 also, I made this crap build again, gonna deploy it without testing because YOLO Oct 08 20:29:12 pretty sure I'm missing a bunch of packages too, but nevermind Oct 08 20:29:15 BOOOOOM! Oct 08 20:30:17 -rw-r--r-- 1 jwh jwh 737 Oct 8 16:18 01_cmake_sucks.patch Oct 08 20:30:17 -rw-r--r-- 1 jwh jwh 2630 Oct 8 17:59 02_musl-fixes.patch Oct 08 20:30:17 -rw-r--r-- 1 jwh jwh 14089 Oct 8 17:52 05_musl_sucks.patch Oct 08 20:30:19 :D Oct 08 20:33:58 well I'm about to dog-food my commit.....and update some qnap firmware. I may be some time. Oct 08 20:34:04 o/ Oct 08 20:34:26 gonna cut my hair and then deploy untested beta builds on production firmware and then go to sleep :D Oct 08 21:14:53 alright, I'm going to send my mesongx target Oct 08 21:14:59 this has been sitting in my tree for too long Oct 08 21:16:15 i wanna adapt that to support other s905 boards one day Oct 08 21:16:47 once this is in master, adding a new board should be fairly easy Oct 08 21:17:19 yeah but the odroid c2 uses a different u-boot/image format than everything else Oct 08 21:47:28 hah, damn Oct 08 22:00:56 ARMv8 servers with UEFI and ACPI should be ...just... around the corner, like last year - and the one before or the year before that Oct 08 22:01:11 ha! Oct 08 22:03:40 hehe Oct 08 22:34:44 its ok tho most s905 devices also have onboard emmc as well as a sd card slot so there's also all that Oct 08 22:43:34 yeah I don't have an eMMC module so have not tested that Oct 08 22:44:42 initially I made mesongx target with gxbb subtarget, but I decided to drop the subtarget. the main target should be good to support all gx variants Oct 08 22:44:48 maybe even axg Oct 08 22:45:22 wilson3468: do you have s9* devices ? Oct 08 22:58:42 sigh. keep ending up with an image with old stuff, so there's still something wrong Oct 08 22:58:50 then it won't be for tonight :( Oct 08 23:00:06 hm Oct 08 23:00:28 does fw3 have ebtables support? Oct 08 23:00:55 oooor, does netifd understand fancy bridge filters? Oct 08 23:01:06 and/or vlan aware bridges? Oct 08 23:27:22 hmph, where are my sysupgrade images Oct 08 23:32:19 must have disabled something that they depend on Oct 09 00:16:28 so it seems that in 18.02 your missing comp_lzo in /usr/share/openvpn/openvpn.options Oct 09 00:24:21 how does /etc/config/network get synthesized? Oct 09 00:31:49 philipp64|laptop: depends on the target, but e.g. target/linux/ar71xx/base-files/etc/board.d/02_network **** ENDING LOGGING AT Tue Oct 09 02:59:59 2018