**** BEGIN LOGGING AT Wed Oct 17 03:00:00 2018 Oct 17 04:07:01 ugh Oct 17 04:07:08 thought it was ebt Oct 17 04:07:12 but apparently not Oct 17 04:07:21 something in 4.14 is causing a panic that didn't in 4.9 Oct 17 04:07:58 ebt? Oct 17 04:08:04 ebtables Oct 17 04:08:15 even now that shit causes constant panics Oct 17 04:09:03 I don't think it was ebt after all though Oct 17 04:09:06 its still doing it Oct 17 04:09:13 just taking longer Oct 17 04:09:29 took 15 hrs but it crashed Oct 17 04:09:40 same config as 4.9 Oct 17 04:10:37 mvebu? Oct 17 04:10:39 x86 Oct 17 04:10:54 max load of maybe 3gbps Oct 17 04:11:07 out of the 34g it has Oct 17 04:11:11 were you monitoring the serial port? Oct 17 04:11:50 no serial unfortuantely, apparently sun didn't bother making it work on this ilo release Oct 17 04:11:56 unfortunately* Oct 17 04:12:05 I need to fix the coredump mask Oct 17 04:12:11 hope its not usb thats breaking Oct 17 04:12:18 as its usb booted Oct 17 04:12:43 i am connected through usb-booted x86 atm, no panics yet Oct 17 04:12:50 hm Oct 17 04:13:07 I have a bunch of extra kernel stuff enabled but nothing thats even in use Oct 17 04:13:14 or would cause a panic without being explicitly used Oct 17 04:14:05 and it should be the kernel that makes serial work Oct 17 04:14:54 I think the serial works fine, its just the ilo thats broken Oct 17 04:15:50 its really annoying in that it always crashes in the middle of the bastard day Oct 17 04:15:59 when it needs to work Oct 17 04:26:06 if serial works when the kernel is booted, you can monitor serial console Oct 17 04:51:41 'lo Oct 17 05:40:37 DonkeyHotei: well thats the problem, serial console (emulated, connected to the BMC) appears to work, but the BMC doen't display anything Oct 17 05:40:49 so all I can get is a crash dump really Oct 17 06:23:04 dedeckeh: ping Oct 17 06:24:49 dedeckeh: regarding the "changed DUID not applied on odhcpd reload" issue - you said a while back that you found an issue related to it and fixed it, do you remember the specific commit? Oct 17 06:35:21 jow:I did not fix it yet as it needs a rework of the static leases reload logic Oct 17 06:35:59 jow:Unfortunately it's not a trivial bugfix when I had a deeper look into the problem Oct 17 06:43:32 jow:regarding moving ip-tiny/ip-full to /usr/libexec I'm wondering if the install section for the variants need to create the install dir /sbin ? Oct 17 06:44:29 jow:the patch I've now is https://git.openwrt.org/?p=openwrt/staging/dedeckeh.git;a=commitdiff;h=710d12e49a69a8121b98d994db07e48f15d5d3cb Oct 17 07:06:47 dedeckeh: since https://git.openwrt.org/?p=project/opkg-lede.git;a=commitdiff;h=52fc0069b50b9be89ae3b525b98741123461dd0a;hp=c668fcec731bed7acc21b3f88f9a94cb55b9b498 opkg will take care of creating the parent dirs for the alternatives symlink Oct 17 07:09:55 jow:ah I was unaware of that;thx Oct 17 07:46:32 oh I remember filing bugs a while ago hmm time to check if anything happened at all Oct 17 07:51:18 found it https://github.com/openwrt/luci/issues/2143 I guess I should have used openwrt bug tracker instead ww Oct 17 07:52:30 oh wait no the bug tracker says to use github Oct 17 08:02:38 Xiatian: it was being discussed here about an hour ago. It is on the radar. Oct 17 08:03:05 makes sense Oct 17 08:28:42 Xiatian: 08:35 < dedeckeh> jow:I did not fix it yet as it needs a rework of the static leases reload logic Oct 17 08:28:47 Ycarus: 08:35 < dedeckeh> jow:Unfortunately it's not a trivial bugfix when I had a deeper look into the problem Oct 17 08:28:58 sorry Xiatian ^ Oct 17 08:29:04 also its not luci related apparently Oct 17 08:29:10 oh huh Oct 17 08:29:22 thanks for the heads-up Oct 17 08:29:34 so the fix will most likely happen in odhcpd Oct 17 08:34:23 https://github.com/openwrt/packages/pull/7209 - hoping for a review Oct 17 08:37:00 have been reading http://www.sartura.hr/blog/first-contributions-towards-integrating-4.19-kernel-in-openwrt/ Oct 17 08:37:43 wondering what is meant by "OpenWrt's flexibility in this regard is not up to standard of some other projects, and we were in some places forced to patch the OpenWrt source itself to make things work. " especially the first part of that sentence. Oct 17 08:40:04 it means "Satura would like to continue making money off OpenWrt. They already sold network accelleration features to customers and now need to merge Kernel 4.19 ASAP so that the community takes care of maintaining it for free" Oct 17 08:40:36 or something along these lines ;) Oct 17 08:42:33 "OpenWrt's flexibility in this regard is not up to standard of some other projects" - theyse guys continue to refuse the patches we need Oct 17 08:43:13 * ldir detects that jow is in 'subtle' mode today Oct 17 08:44:02 seriously, every single contribution from anyone @sartura is for their sole self interest Oct 17 08:44:17 they give a flying shit about the openwrt ecosystem Oct 17 08:50:49 noted Oct 17 08:51:20 is there any reason why we don't generate vendor-compatible firmware image for Ubiquiti EdgeRouter X? Oct 17 08:51:22 so it could be flashed just like official firmware, using some vendor web UI? Oct 17 08:51:49 vendor seems to be using simply a tar archive with vmlinux + rootfs Oct 17 08:51:50 xback: ping Oct 17 08:51:58 ping russell-- ^^ Oct 17 08:52:15 ldir: its just my personal opinion anyway, take it with a grain of salt Oct 17 08:55:45 jow: don't worry - if any consolation there were a few points in that blog post (and others) that made me feel uncomfortable. Oct 17 08:57:40 ping blogic, you are the mainteiner ;) maybe you can answer me that Oct 17 08:57:48 blogic: can you see my above question, please Oct 17 08:58:53 I've 'questions' about prpl as well..... there's a lot for me to discover at the conference :-) Oct 17 08:58:59 jow: pong Oct 17 08:59:55 xback: in fec8fe806963c96a6506c2aebc3572d3a11f285f ("kernel: bump 4.9 to 4.9.116") you dropped patch 403-mtd_fix_cfi_cmdset_0002_status_check.patch since it allegedly has been upstreamed Oct 17 09:00:39 jow: checking .. Oct 17 09:00:40 xback: the removal of this patch reportedly causes issues on at least the wzr-hp-g300nh (an ar71xx device with very large nor flash) Oct 17 09:01:16 the problems manifest in random I/O errors on jffs2 corruptions Oct 17 09:01:29 the forum thread tracking it is at https://forum.openwrt.org/t/impossible-to-install-update-any-packages-on-wzr-hp-g300nh-18-06-1/21666/1 Oct 17 09:02:07 and it got tracked down to the patch removal in post https://forum.openwrt.org/t/impossible-to-install-update-any-packages-on-wzr-hp-g300nh-18-06-1/21666/119 Oct 17 09:02:30 could it be that the patch was only partly or differently upstreamed? Oct 17 09:03:39 looking at the code again, hold on Oct 17 09:11:11 rmilecki: pong Oct 17 09:11:30 russell--: did you see my question Oct 17 09:11:31 ? Oct 17 09:11:39 ah, yeah. i don't know. Oct 17 09:11:50 * russell-- flashed from an initramfs Oct 17 09:12:11 ok Oct 17 09:12:24 it is super easy to open the case ;-) Oct 17 09:12:37 two screws Oct 17 09:12:51 i just happen to often care about user experience ;) Oct 17 09:12:56 but yeah, a "factory" image would be nice for most people Oct 17 09:13:29 jow: the removal is due to upstream commit https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.9.133&id=a0239d83e1cb60de5e78452d4708c083b9e3dcbe Oct 17 09:13:47 It fixes partial and full erase Oct 17 09:14:52 Looking at our patch, it seemed it was needed due to malfunctioning of these erase functions, so I also removed it (and tested on ar71xx with NOR) Oct 17 09:14:57 you should only be able to download an openwrt image by uploading a selfie while holding a 3.3V usb-serial adapter and a local newspaper for proof Oct 17 09:15:14 I didnt see any fallout during testing it Oct 17 09:16:00 I'll make a patch which restores the chip checks for the write operations, and push it in my staging tree so people in the thread can test it Oct 17 09:17:39 russell--: i have zero experience with U-Boot Oct 17 09:17:42 i just read "The resulting bin/ramips/openwrt-ramips-mt7621-ubnt-erx-initramfs-kernel.bin file can be booted via option 1 in the boot menu available via a serial connection." Oct 17 09:17:52 russell--: do you mind giving me a very quick help on that? Oct 17 09:18:03 you need a tftp server Oct 17 09:18:10 russell--: how do I access U-Boot menu? Oct 17 09:18:15 serial console Oct 17 09:18:22 right... and then what? Oct 17 09:18:31 should I press some key while it boots? Oct 17 09:18:35 i think ctrl-C Oct 17 09:18:37 during boot Oct 17 09:18:38 yeah Oct 17 09:18:39 ok sounds good Oct 17 09:18:41 thanks Oct 17 09:18:48 my unit should arrive in an hour or so Oct 17 09:18:49 or maybe just the 1-key Oct 17 09:19:01 i just did it, i should remember Oct 17 09:19:11 i think just start pressing "1" Oct 17 09:19:47 it'll ask you a few things, including an ipaddr for your device, a server ip address and a filename on the tftp server Oct 17 09:20:52 rmilecki: btw, this "press a number" thing is a ralink(/mediatek?) u-boot only thing, and not found for other u-boots - those have a normal console (on serial) by default Oct 17 09:20:55 i had ethernet in the left most port. that might not matter for tftp, but it's handy once the thing has booted. Oct 17 09:21:07 KanjiMonster: ok, th Oct 17 09:21:09 x Oct 17 09:21:10 57600 baud too Oct 17 09:21:21 which is also a ramips thing Oct 17 09:22:26 if you want to poke around at a u-boot console first, you can use option 4 Oct 17 09:22:41 iirc, 3 is the default, booting from flash Oct 17 09:28:15 rmilecki: let me know if you need the serial console pinout, i'm looking at a photo i took to document it Oct 17 09:28:35 russell--: i found photo on wiki Oct 17 09:29:02 https://wiki.openwrt.org/toh/ubiquiti/ubiquiti_edgerouter_x_er-x_ka#photos Oct 17 09:29:37 yeah Oct 17 09:29:39 xback: thanks! Oct 17 09:29:40 just found it too Oct 17 09:30:23 jow: no problem. sorry for the fallout .. I try hard to test as much as possible .. :( Oct 17 09:32:18 xback: I think you do a bloody wonderful job - and the odd bit of fallout just proves to me you're not a robot ;-) Oct 17 09:41:34 ldir: thanks Oct 17 10:00:51 xback: I'd be interested to know your 'bump kernel' routine and hardware you build on. Oct 17 10:02:08 ldir: in short, I refresh all versions at once and distile the patches from it Oct 17 10:02:28 xback: no reason to be sorry Oct 17 10:02:36 ldir: then I build 8 different targets at once using the bump Oct 17 10:02:55 ldir: i'm spoiled with a 40-core machine with 384GB RAM, of which 196GB is a ramdrive .. Oct 17 10:03:33 ldir: so it takes about 20 minutes to build 8 targets in parallel from scratch Oct 17 10:03:52 bloody hell! :-) Oct 17 10:04:14 ldir: then I have some scripts which extracts all binaries, and flashes it automatically to 8 different boards here on my desk, and to about 20 devices in the field Oct 17 10:04:26 ldir: those devices all run a custom made mesh protocol in userspace Oct 17 10:05:58 wow! This explains why I struggle on my dual core macbook :-) Oct 17 10:06:20 * russell-- only has a 16-core build machine, jealous! Oct 17 10:06:28 ldir: if these devices last long enough without a hickup, I flash the testbuilds to the field, roughly about 40 devices (6 different hardware types) Oct 17 10:07:07 xback: what is the mesh protocol? Oct 17 10:07:29 have you done any sort of presentation/talk on your setup?..... hint! ;-) Oct 17 10:07:41 russel--: i took RFC3626, and fully modified it to support Wifi RSSI bidirectionally Oct 17 10:08:24 interesting Oct 17 10:08:34 ill share a pic. sec Oct 17 10:08:38 * russell-- recalls olsrd people talking about different metric Oct 17 10:09:44 olsrd2 does some link layer aware metric iirc Oct 17 10:10:12 but its been ages that I looked, it seemed to suffer from a severe case of second systems syndrome Oct 17 10:10:50 Here is the output of my binary, just took it live from the windfarm: https://pastebin.com/eXWbgagM Oct 17 10:11:11 all the output is shared through custom olsr messaging Oct 17 10:11:54 our wireless meshes are small enough that batman-adv does the trick for us. Oct 17 10:12:47 the "-1" rssi indicates the link is wired, which is preferred over wifi Oct 17 10:13:31 this allows to do OLSR over multiple techs, ethernet, wifi, bluetooth, LTE .. and favour the usage of each tech Oct 17 10:13:33 * russell-- worked for an oceanographic institution for a while that was using olsrd to handle some instrument telemetry network, i spent a week on a research cruise war-sailing their network Oct 17 10:14:02 ldir: hope this answers your question a bit :) Oct 17 10:15:25 maybe he got scared ;) Oct 17 10:17:06 The main advantage for me is the access to a lot of different hardware, allowing mass testing of code Oct 17 10:17:41 as all devices are moving and using wifi, it allows to test mac80211 updates deeply Oct 17 10:21:05 nice line of sight from the wind turbine towers! Oct 17 10:25:32 Visual representation from the webUI: http://www.xback.be/norther.png Oct 17 10:30:03 and to close of .. I also made a tool which translates all device reports (every 5 sec) in earth live: http://www.xback.be/earth_live.jpg Oct 17 10:43:13 nice Oct 17 10:51:57 i was contacted by someone from a dredging company in antwerp a few years ago, doing some kind of ship-to-ship mesh, possibly using 802.11s. Oct 17 10:52:46 could be DEME? they dredge a lot in the port over there Oct 17 10:53:09 yeah, i think it was Oct 17 10:53:23 i offered to come help them sort it out ;-) Oct 17 10:55:41 We are still looking for an embedded dev here ;) Oct 17 12:18:16 russell--: oh, there IS openwrt-ramips-mt7621-ubnt-erx-initramfs-factory.tar Oct 17 12:18:24 and I just flashed it using verndor's web UI Oct 17 12:18:27 nice! Oct 17 12:26:56 russell--: root@OpenWrt:/etc/config# hexdump -C dropbear Oct 17 12:26:58 00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| Oct 17 12:27:00 finally... i got it Oct 17 12:31:06 got a way to reproduce it reliably? Oct 17 12:38:34 nbd: i bought ramips device Oct 17 12:39:00 russell--: i did few tests just to be sure, I can reproduce it every time Oct 17 12:39:16 compiling 4.9 based image now & going for a dinner Oct 17 12:39:20 tomorrow I'll be hacking on that Oct 17 12:42:00 jow: fix pushed to my staging for the mtd issue. Requested them to test it .. Oct 17 13:12:54 ldir: ping Oct 17 13:13:12 xback: pong Oct 17 13:14:56 ldir: Any idea what dependency could be missing here for wireguard? https://forum.openwrt.org/t/impossible-to-install-update-any-packages-on-wzr-hp-g300nh-18-06-1/21666/135 Oct 17 13:16:22 xback: thats the usual error that happens when one attempts to install repo kmods on custom builds Oct 17 13:16:46 xback: ^^^^ jow just responded as I looked Oct 17 13:16:47 if the local kernel has been built without all kmods enabled, it'll have a different version magic, making it incompatible Oct 17 13:17:08 aah .. that explains Oct 17 13:18:17 there's no clue in that message though, would be good if it said "magic number mismatch" Oct 17 13:19:26 'cos I looked at "kernel (= 4.9.131-1-324914f1f8afebd4db8abc0254414b72) * kernel (= 4.9.131-1-324914f1f8afebd4db8abc0254414b72) * kernel (= 4.9.131-1-324914f1f8afebd4db8abc0254414b72) *" and went 'huh? those numbers match' :-) Oct 17 13:20:19 ldir: well it would probably help to rework the opkg error reporting Oct 17 13:20:30 by filterign out dupes and reword the message Oct 17 13:21:11 then it'd read like "Cannot satisfy the following dependencies for wireguard: kernel version 4.9.131-1-324914f1f8afebd4db8abc0254414b72" Oct 17 13:21:35 this is another https://pbs.twimg.com/media/DJyAs4rXoAIqyMb.jpg isn't it?! Oct 17 13:22:00 wouldn't say so Oct 17 13:22:14 can take a stab at it later Oct 17 13:23:11 let's just say it depends on your coding skill level ;-) Oct 17 14:01:54 ldir: mind setting your full name in github to avoid merge commits with "Author: Kevin D-B <6500011+ldir-EDB0@users.noreply.github.com>" ? Oct 17 14:06:43 btw, it seems ar71xx/generic fails to build since the switch to 4.14 Oct 17 14:07:12 or rather since shortly before Oct 17 14:07:39 jow: I saw the report regarding kernel size for a few subtargets Oct 17 14:07:55 apparently nbd's 212aa332267641fa9fc9a16d39159decec95b4a1 or 1e8882585c6f4a5e7f5e2b4f18cd550aafa6f81d made the kernel too big to fit for the ens202ext board, even with 4.9 Oct 17 14:09:55 stintel: done - apologies - I hate the github interface anyway.... I knew I should have done those merges longhand. Oct 17 14:10:15 most devices have 1.5MB, there are some with 1.6MB and even those fail as it's too big Oct 17 14:11:16 ldir: no worries, I just noticed it while testing mangix' net-snmp 5.8 bump Oct 17 14:11:24 I think we should make the "too big" fatail by default and a nonfatal warning with CONFIG_BUILDBOT=y Oct 17 14:11:51 that should match everyones expectations Oct 17 14:12:03 ^^^ +1 ! Oct 17 14:12:20 a few targets could also move to the Tiny, so kernels symbols etc get disabled. Would this also be an option? Oct 17 14:14:11 yes, definitely Oct 17 14:14:20 I was waiting a bit to see if kernel 4.19 is smaller than 4.14 as code got dropped along the way Oct 17 14:20:47 FS report about the kernel size: https://bugs.openwrt.org/index.php?do=details&task_id=1891&order=id&sort=desc Oct 17 15:32:44 greearb: hi! just wanted to say that I'm getting this from time to time with the latest beta: [38963.262336] ath10k_pci 0000:01:00.0: Unknown (10.4) eventid: 36925 Oct 17 15:38:55 jow: kernel size fix pushed to my staging --> https://git.openwrt.org/?p=openwrt/staging/xback.git;a=commit;h=93e51a77f1b311cac86f89cc9da2e7e893cb5b50 Oct 17 15:39:06 jow: checking for more broken ones .. Oct 17 15:48:22 hi… I'm working on some build automation for OpenWRT, and was wondering about caching a copy of "tools/compile". I could do a "/prepare" of all of these, then collect their directory checksums, and compare those with the results of a previous build… if anything changed, rebuild everything (because of headers and static libraries having a rippling effect). but otherwise I should be able to re-use the previous result since it' Oct 17 15:48:23 s not changed, right? am I missing anything? Oct 17 15:51:34 huaracheguarache, plz open a bug to track it, I'm knee deep in rate-ctrl code at the moment Oct 17 15:55:17 greearb: Only knee deep? Go on, get in there properly.. :P Oct 17 16:13:36 xback: btw, you can observe the buildbot status at https://phase1.builds.lede-project.org/ Oct 17 16:14:09 jow: thanks! Oct 17 16:15:14 jow: is it possible to get a mail or so automatically on build fail? Oct 17 16:16:19 xback: not yet but it seems to be easily doable; will try to set it up Oct 17 16:17:10 jow: would be nice to send notifications to at least the people defined in SoB etc Oct 17 16:17:30 that would be problematic as we currently coalesce changes Oct 17 16:17:41 so people can get mails abuit totally unrelated failures Oct 17 16:17:52 greearb: will do Oct 17 16:17:58 but will see if I can restrict it to the committers, not the authors then it would be okay I guess Oct 17 16:18:01 Monkeh, maybe just knees sticking out the top :) Oct 17 16:18:47 anyway, got about 30% improvement in 20-station throughput with the 9984 firmware...will upload it soon as it gets a bit more testing and folks can try it out. Will check wave-1 for similar improvements as well. Oct 17 16:19:05 jow: looks more than enough that way. thanks Oct 17 16:49:18 hmmm Oct 17 16:49:48 is there anyway to get the dmesg crash output either kept in ram over a reboot, or dumped to disk? Oct 17 16:51:17 jwh: not that I know of. Best way is probably to hook up serial console and capture it there? Oct 17 16:51:48 yes, there's a menuconfig option to keep it in ram for next boot iirc. Oct 17 16:51:59 karlp: what's that one called? Oct 17 16:52:00 have no working serial console thats the problem :P Oct 17 16:52:17 ilo says its working but I get no output Oct 17 16:52:18 make menuconfig and /crash ? Oct 17 16:52:45 sun boxes, no physical serial attached Oct 17 16:53:09 probably gonna have to roll this back anyway unless it crashes during the night Oct 17 16:54:13 hm whats KERNEL_CRASHLOG? openwrt symbol Oct 17 16:55:52 oh, !x86_64 anyway Oct 17 16:57:00 jow: what's the status of the EFI support merging into master? Oct 17 16:59:53 so uh, if I use the crash logging thinger, do I need something enabled in userland also to read it, or can I just get it from the kmsg buffer after reboot? Oct 17 17:04:25 hate these sorts of bugs, no idea what the trigger is or how long before it gets triggered Oct 17 17:04:32 could be 5 minutes, could be 24 hrs Oct 17 17:05:51 any thoughts about CONFIG_PKG_LTO to globally enable LTO? https://gist.github.com/95c253019bc666f3ebd29dffbe194472 Oct 17 17:08:20 huh, pstore seems like a good idea Oct 17 17:10:35 stintel: if it defaults off, sure. Oct 17 17:11:02 or you know, turn it on and use openwrt devs as guinea pigs to finally get lto out mainstream Oct 17 17:11:14 it'll shake out plnety more bugs in both package code and lto code. Oct 17 17:14:36 karlp: oh it will :) Oct 17 17:16:23 * stintel & Oct 17 17:53:16 philipp64: no change Oct 17 18:11:11 stintel: i like it Oct 17 18:12:39 spesking of LTO, there's a bug with GCC7 where it fails often. Fixed in 7.3.1 Oct 17 18:40:36 stintel: with default off it is good Oct 17 18:51:53 jow: so what needs to happen to unblock it? Oct 17 19:07:19 philipp64: ideally someone who can test it submits as new patch series Oct 17 20:22:01 Evening, could somebody explain to me the the following entry in the OpenVPN tutorial here - https://openwrt.org/docs/guide-user/services/vpn/openssl-pki#install_packages Oct 17 20:22:25 Section: Create Certificates -> CA -> Section 2. Oct 17 20:22:39 What do I need to input in the Server and the Client lines Oct 17 20:51:03 rmilecki: thanks for the confirmation Oct 17 20:51:29 sure Oct 17 20:52:43 esters: I'd think it's the subject for the CA Oct 17 20:53:31 rmilecki: interesting that you didn't see it on bcm53xx Oct 17 20:56:33 russell--: hopefully debuging will tell us why Oct 17 20:56:38 it's weird indeed Oct 17 22:18:40 ldir: can you add PKG_CPE_ID:=cpe:/a:libssh:libssh to the CVE PR so I can kill it locally? Oct 17 22:21:16 https://github.com/openwrt/packages/commit/1779f617e65e3c7efaf664690d640d595777dc4d#diff-4ca7da4886449d464d5080f58667e25eR21 is an odd ball given everything else is using CPE 2.2 Oct 17 22:23:12 swalker: done Oct 17 22:31:12 local list is https://pastebin.com/raw/ZCiWg7PA if there's any takers Oct 17 22:37:23 huaracheguarache, greearb: looks like this https//git.openwrt.org/a330ecebf57e71bb489ccf8e5f049ccf982e1bfc might be needed for ath10k-ct as well Oct 17 22:38:25 swalker: just added c-ares Oct 17 22:39:24 I don't see anything else in your list that I look after. Oct 17 23:09:45 karlp: the c-ares: before the cpe string needs removed, it's only for package matching in my local script Oct 18 00:08:49 what would cause mtd_read error during early boot? Oct 18 00:20:53 swalker: not sure I follow. You want more of the cpe strings added to the packages right? I just added it to the c-ares package Oct 18 00:20:56 from offset 0x3AE0000 to 0x7FE0000, my personal guess would be the nand is mapped incorrectly in memory... Oct 18 00:40:30 karlp: it needs to be PKG_CPE_ID:=cpe:/a:daniel_stenberg:c-ares rather than PKG_CPE_ID:=c-ares:cpe:/a:daniel_stenberg:c-ares Oct 18 00:48:53 rmilecki: now that ar71xx has bumped to 4.14, i can test my rb493g Oct 18 02:06:09 Dear Compex: Shorting two power rails together is not kosher, even if they are the same voltage. Oct 18 02:26:45 * m4t is curious if there's a performance difference in the same boards ath79 vs ar71xx 4.14 Oct 18 02:57:56 rmilecki: ar71xx/rb493g is also affected **** ENDING LOGGING AT Thu Oct 18 03:00:00 2018