**** BEGIN LOGGING AT Mon Apr 29 02:59:56 2019 Apr 29 06:04:11 nbd: in principle fine, the naming is not ideal though Apr 29 06:04:32 testing kernel somehow implies a testing kernel.org version Apr 29 06:04:40 can't think of a better name right now though Apr 29 06:07:15 nbd: maybe one of bleeding edge kernel, unstable kernel or upcoming kernel Apr 29 06:09:12 I like upcoming Apr 29 06:09:51 could you please tell the status of tp-link archer c7 v1? Apr 29 06:11:03 which status? Apr 29 06:11:29 if the boot-loop for ath10k is fixed. on the mailing-list there is some talk about this issue Apr 29 06:12:04 if the discussion happens there, maybe its the best place to look for a solution? Apr 29 06:12:18 havent seen anything about this here Apr 29 06:17:56 rubberduck: I think there's no change, you still have to build a custom image without kmod-ath10k included Apr 29 06:18:04 rubberduck: i believe it's already fixed. Apr 29 06:18:13 * mangix looks Apr 29 06:21:08 https://patchwork.ozlabs.org/patch/1073218/ Apr 29 06:21:13 set as rejected Apr 29 06:21:26 so no, not fixed. Apr 29 06:22:12 oh this is for ath10k, not -ct. wouldn't fix it anyway. Apr 29 06:24:09 i have a c7-v2 which is working fine and a c7-v1 which doesn't work at 5 GHz. Apr 29 06:24:31 i hope this one will be supported soon to avoid electronic-waste. Apr 29 06:24:37 regarding 5ghz the device are really different Apr 29 06:24:43 *deviceS Apr 29 06:24:57 rubberduck: the v1 has a removable pci card. Apr 29 06:24:57 v2/v4/v5 are more similar than v1 Apr 29 06:24:58 only one letter *g* - A instead of B on the WIFI Card Type Apr 29 06:25:26 well, ask the vendors... Apr 29 06:25:45 the A variant will never be supported Apr 29 06:26:13 The only thing that can be done is to not have ath10k panic Apr 29 06:26:21 ok - so i should buy a v2 or even better a Fritzbox 4040 Apr 29 06:26:32 or replace the pcie card Apr 29 06:27:08 as i read the mailing-list there is also an issue with the pcie slot with the v1 so replacing maybe won't work, too Apr 29 06:27:13 restore stock fw on the v1 and sell it on ebay, that way its not wasted Apr 29 06:27:43 what problem would that be? Apr 29 06:28:39 newer openwrt are enabling both pcie ports of the SOC but the 'first' is not populated Apr 29 06:29:03 that has nothing to do with whether or not the pcie port on the v1 works Apr 29 07:10:00 Is there some general guidance as to how much of NAND to leave unallocated to MTD to allow for bad blocks? Apr 29 07:10:35 rubberduck: that bootloop has been fixed by removing the ath10k kernel modules in master git.openwrt.org/34113999ef and in 18.06.y git.openwrt.org/9c4fa1bb Apr 29 07:15:19 there's ongoing attempt for the proper fix https://patchwork.kernel.org/patch/10866417/ Apr 29 07:16:39 BTW I just assume, that it has been fixed, but I didn't get any feedback from any of the affected users yet, so who knows Apr 29 09:23:55 * KevinDB is now running on his apu2 :-) Apr 29 09:24:20 ah Apr 29 09:25:09 * ldir assumes correct identity :-) Apr 29 09:26:05 :) Apr 29 09:28:34 the cpus are going to be VERY bored. Apr 29 09:33:36 ldir: run surricata or so :) Apr 29 09:34:19 ldir: You can always load it up with transmission, three VPNs, samba on NTFS, minecraft, ... Apr 29 09:35:33 and pi-hole in an lxc container ;) Apr 29 09:36:42 ...and why can't I just build packages on my router?? Apr 29 09:37:08 yes, it's currently only doing what the old archer c7 v2 was doing...which ain't much :-) Then I went to the mir3g for a few days...that was step up in that '4 cpus!' - I could have more than got away with just using the mir3g. Apr 29 09:37:37 but shiny new x86 toy! :-) Apr 29 09:38:05 could do a HA setup now ;) Apr 29 09:38:26 Like mine a lot -- especially at 7 W Apr 29 09:38:51 Time to port ZFS to OpenWrt, I think ;) Apr 29 09:40:02 i'd rather spend my time porting bcachefs and testing that and reporting any encountered bugs Apr 29 09:41:05 ZFS as openwrt FS (on hardware capable of) would tickle my fancy too :p Apr 29 09:41:33 But at the same time I do see it not being the main point (: Apr 29 09:41:45 lols at memory stats - 4GB free 100M used (rounded up bigly) Apr 29 09:42:31 It shouldn't be too bad -- I've got Debian running on my APU3C4 on ZFS with mirrored SSDs, but that's probably more than most OpenWrt users would go for Apr 29 09:44:07 Now if I could only see inside of this chip and figure out why the spi-nand driver (WIP) thinks I have 992 bad blocks Apr 29 09:46:28 Well I mean simple version, in theory, is to have one more option in buildroot menuconfig, but as with ZFS has all the cool stuff baked in like pools and such there is propably never too clean way of doing gazillion variations of root pools and so on.. when going to the other extreme ;) Apr 29 09:46:50 I enabled the hardware watchdog and proved that works - link pins 5&6 of J2 Apr 29 09:52:17 Tricky thing with ZFS is that the license isn't compatible with GPL -- basically has to be built by the Linux end user, as I understand it Apr 29 09:53:44 At least we have buildroot to do that ;) but yeah.. Apr 29 09:54:23 "But I want to run it on my 15-year-old TP-Link..." Apr 29 09:54:27 ZFS on a Router? Apr 29 09:55:00 Yep, ldir has new toys and is complaining that they are underutilized Apr 29 09:56:08 really? Apr 29 09:57:08 Everything else at my posession runs ZFS pretty much, so why not router too? Again I say this with no real hopes it happening but it should ;) Apr 29 09:57:56 Also again, on devices suitable to use ZFS, so not raw flash devices obviously Apr 29 09:57:57 sorry but nearly every affordable router has not enough RAM for running ZFS Apr 29 09:58:01 Well, APU2/3C4 is a quad-core x86 with 4 GB of memory and it happily will run ZFS under FreeBSD or Debian Apr 29 09:58:24 jeffsf: you may run this devices with 'normal' linux, too... Apr 29 09:59:04 So mainly x86 and then these purpose built stuff like APU that was just mentioned Apr 29 09:59:04 Yep, just giving him something to do with all those spare resources :wink: Apr 29 09:59:28 jeffsf: install 'normal' linux and openwrt as a VM on top (kvm/xen) Apr 29 09:59:42 Or just install Debian Apr 29 09:59:54 this is a "normal" Linux Apr 29 10:00:13 since I don't need the wireless support on that box Apr 29 10:00:55 At our hackerspace we use openwrt in some Avoton c2750 as router, because openwrt just shines in this task Apr 29 10:00:56 I agree, ZFS isn't really an OpenWrt thing. I'd rather see nftables than ZFS, personally Apr 29 10:01:15 Yep Apr 29 10:10:31 stintel: hmmm.... ccp 0000:00:08.0: BAR 0: can't reserve [mem 0xfeb00000-0xfeb1ffff] ???? Apr 29 10:11:48 ldir: no idea :) Apr 29 10:12:06 ok, good :-) Apr 29 10:12:21 ldir: but afaik it works Apr 29 10:14:36 this always happens - ldir gets new toy - pokes at bits of it going 'wot that do?' Apr 29 10:18:51 yeah, I remember adding those kmods Apr 29 10:19:16 but they couldn't be loaded at that time because no IOMMU support in the BIOS afaik Apr 29 11:04:18 fyi, new kernel bumps pushed to staging Apr 29 11:15:01 I've appended default packages in my targets Makefile to include some kmods. After doing so I can see in .config CONFIG_DEFAULT_kmod-xyz=y however CONFIG_PACKAGE_kmod-xyz is not set. What is the purpose of the CONFIG_DEFAULT? I assumed it was just used to set the CONFIG_PACKAGE. Thanks Apr 29 11:16:25 Btw I've run make menuconfig after making my changes. I thought that would sync up any issues Apr 29 11:17:43 Neighbor11111114: afaik it is only used when creating a new config and existing configs will not be affected Apr 29 11:18:24 Ahh I see thanks stintel, do you what the purpose of the CONFIG_DEFAULT_xyz is then? Apr 29 11:19:05 well that those packages are enabled when creating a new config, but allowing the user to disable those packages if they are not needed Apr 29 11:19:22 (I think) Apr 29 11:21:50 ahh I see ok thanks. I moved .config to .config.backup and tried menuconfig and it got recreated. Thanks for the assitance :) Apr 29 11:22:36 yw Apr 29 17:40:04 Is CONFIG_MTD_NAND_AR934X purely an ar71xx thing? Apr 29 17:41:56 (and would it have been required for an qca9563) Apr 29 18:39:55 I have a question about compiling host packages in Openwrt Apr 29 18:43:01 brucethompson: what is your question? Apr 29 18:43:34 First, is there a specific make target for host packages? Apr 29 18:44:09 afaik, yes, typically Apr 29 18:44:59 What is it? In the docs, it talks about {/download, /compile, /install} Apr 29 18:45:04 brucethompson: which host package are we talking about? Apr 29 18:45:13 Libsodium Apr 29 18:45:41 I have a dependancy on that and it doesn't seem like the host package is being built Apr 29 18:48:48 it is in the main tree or in a feed? Apr 29 18:49:28 packages feed Apr 29 18:50:16 Not sure what your question is. Apr 29 18:50:22 I am using the SDK. Apr 29 18:50:30 Came from the main feed Apr 29 18:50:42 Very recent Apr 29 18:52:19 i'm not sure what your question is either ;-) Apr 29 18:53:01 I am doing a make of a new package. Apr 29 18:53:43 more specifically, i don't know what your answer is. looking at the package Makefile for libsodium in the packages feed, i don't see a host component Apr 29 18:53:54 It is looking in staging_dir/{host,hostpkg} for libsodium.a and libsodium headers Apr 29 18:54:07 They are not there. Apr 29 18:54:20 OK. That's likely the problem Apr 29 18:55:11 the cross compiler should build the library for the target and make the headers available for the dependent packages Apr 29 18:55:26 It's a new package. Apr 29 18:55:30 I wrote the Makefile Apr 29 18:56:08 i don't use the SDK, just the main source tree, so i'm not well equipped to see where you are going wrong Apr 29 18:56:21 Same happens in main source tree Apr 29 18:57:13 OK. Sounds like there are currently no packages in OpenWrt that depend on host version of Libsodium. Apr 29 18:57:14 i'd look in the packages feeds for other packages that depend on libsodium Apr 29 18:57:29 And see what they are doing?? Apr 29 18:57:42 cd feeds/packages ; git grep libsodium Apr 29 18:58:04 Understand. Then check what they did in Makefile, correct? Apr 29 18:58:29 yes Apr 29 18:58:35 OK. I will try that. Thx Apr 29 18:58:55 Have you a line like "DEPENDS+=+libsodium" in your Makefile ? Apr 29 18:59:12 see libs/zmq/Makefile Apr 29 18:59:21 OK Apr 29 19:00:02 and a few others Apr 29 19:01:09 also see https://openwrt.org/docs/guide-developer/helloworld/start as a useful intro to the build system and package Makefiles Apr 29 19:01:28 Did that a long time ago. Apr 29 19:01:58 Python packages seem to be infinitely more complex than what's in that example. Apr 29 19:02:44 But I followed that for setting up build environment for latest github root Apr 29 19:04:15 then I shall refrain from continuing to teach you how to suck eggs :-) Apr 29 19:04:25 Upgrading nfs-kernel-server on root from 2.3.1-1 to 2.3.3-2... Apr 29 19:04:26 Downloading http://downloads.openwrt.org/releases/18.06.2/packages/mips_24kc/packages/nfs-kernel-server_2.3.3-2_mips_24kc.ipk Apr 29 19:04:28 Collected errors: Apr 29 19:04:30 * satisfy_dependencies_for: Cannot satisfy the following dependencies for nfs-kernel-server: Apr 29 19:04:32 * rpcbind Apr 29 19:04:35 whats wrong? Apr 29 19:04:59 I'm looking at the Makefile you referenced Apr 29 19:05:30 hsp: the message said it can't satisfy the dependency on rpcbind Apr 29 19:06:09 rpcbind is not available Apr 29 19:10:42 OK. The libsodium dependency seems to be slightly different Apr 29 21:24:28 g'nite all **** ENDING LOGGING AT Tue Apr 30 02:59:57 2019