**** BEGIN LOGGING AT Fri Feb 08 02:59:56 2019 Feb 08 06:33:43 rmilecki: I have a Linksys WRT610N v1 (BCM4705+BCM53115+BCM4322), if you still need testing. I could take a look at it in ~12-14 hours Feb 08 06:34:09 pkgadd: great, thanks, let me finish my brcmfmac patch & get to tha Feb 08 06:35:16 o.k., take your time - I'm currently at work/ away from the wrt610nv1 anyways Feb 08 09:49:14 karlp: BTW, can you rule out filesystem? Feb 08 09:50:04 ynezz: there were config differences too yet, I'm doign some more tests today. Feb 08 09:50:27 ran out of night :) Feb 08 10:10:35 Is it possible/how to create a custom firewall chain within the fw3 syntax structure? e.g create in the mangle table a new chain MY_QOS_MANGLE_CHAIN ? I don't think it is but maybe I'm missing something. Feb 08 10:17:39 * ldir giggles. Subconscious types 'iptangles' instead of 'iptables'. I think this clearly shows my mind state :-) Feb 08 10:17:58 hehe Feb 08 10:18:12 I had some point where I was constantly mistyping strongswan as strongsean Feb 08 10:18:40 lucky I don't know any Sean, so there's no need to worry 😂 Feb 08 10:19:01 lol Feb 08 10:55:45 * karlp heavily suggests flock differences. Feb 08 10:57:46 karlp: staging_dir/host/bin/flock ? Feb 08 11:33:55 xback: BTW I've noticed, that you've closed the 4.19 PR for imx6 so I've tried imx6 on 4.19.19 today and here is the result http://paste.ubuntu.com/p/GCk4fXz4F9/ (still missing bunch(9) of DRM symbols) :-) Feb 08 11:34:38 * ldir has done a good deed for the day - edited a small bit of the firewall3 wiki Feb 08 11:48:32 ynezz: no worries, I have the patch locally here. It needed some alteration to exclude the renamed symbol. I just didn't push it to my staging yet Feb 08 11:52:52 anyone know why gtk2.0-dev is listed as a build requiremente for the host? https://openwrt.org/docs/guide-developer/build-system/install-buildsystem#table_of_known_prerequisites Feb 08 11:53:07 or sdcc? Feb 08 11:53:29 jikes too, who put that on the list of requirements? Feb 08 11:56:46 karlp: possibly former dependencies that weren't removed when the tools/packages needing them were updated to newer version Feb 08 11:57:10 I do remember needing to install sdcc in the past Feb 08 11:57:49 the gtk-dependency might be from the time the x11-feed was still a default feed Feb 08 11:58:00 or xorg or whatever it was called Feb 08 11:58:00 yeah, trying to run down this rebuild slowness and it's "just" "build-essential unzip python2" Feb 08 11:58:10 even then, why would you need gtk for the host? Feb 08 11:58:40 I'm inclined to excise heavily from teh list and let things be re-added if people find they need them again. Feb 08 11:59:06 there's this big list of packages for various distros too, but they're often, "this is what was needed when that distro was cutting edge, to build the openwrt tree from that time period" Feb 08 11:59:15 which is... a little different. Feb 08 12:35:42 could you, please, take a look at https://bugs.openwrt.org/index.php?do=details&task_id=2108 ? Feb 08 12:35:53 i even have patch... Feb 08 12:38:16 raorn: that does not belong in bugs.openwrt.org Feb 08 12:38:36 raorn: the correct way is to wait until someone handles the PR Feb 08 12:40:38 raorn: you failed to mention the package maintainer in your PR message, as shown in the PR template Feb 08 12:48:14 are there PR templates? Feb 08 12:53:03 stintel: where do i submit bugs to packages in packages feed? Feb 08 12:53:39 if you have a PR that fixes it, you don't *need* to create a bug per se Feb 08 12:53:56 the template is the text that is in the description at the time you create the PR Feb 08 12:54:24 anyway, I've mentioned the maintainer for you Feb 08 12:54:40 but please do not remove the stuff in the template next time, it's there for a reason Feb 08 12:59:11 oh, now I see. by bad, sorry... Feb 08 13:00:59 np Feb 08 16:47:34 ynezz: ping Feb 08 16:52:48 ynezz: fyi, after your fix for ath79 ethernet irq. I tried it for fun on ar71xx - rocket M5 Feb 08 16:53:31 after a few times downing/upping eth0, I can easily simulate the interface not transferring packets anymore. Feb 08 16:54:34 lol Feb 08 16:54:36 er Feb 08 16:54:59 mistaken tmux keyboard mashing there, nothing to see here, etc. Feb 08 17:25:20 * blogic looks at https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/log/?h=ath11k-bringup and grabs the code review axe Feb 08 17:34:53 43k lines?! :) Feb 08 17:34:57 i know Feb 08 17:35:06 a fw blob and only 11ac suppor Feb 08 17:35:08 +t Feb 08 17:35:17 karlp: might aswell get cracking early though Feb 08 17:37:36 kinda nice seeing ISC licensed code though... Feb 08 17:38:48 maybe the people who wrote it are paid by the line? Feb 08 19:30:20 anyone has an idea what might cause the build system to no longer generate sysupgrade.bin images? Feb 08 19:32:08 valku: your image got too big? Feb 08 19:32:23 checking the size... Feb 08 19:37:22 valku: make V=s will tell you about any size problems. make without V=s will just silently fail Feb 08 19:37:31 (and it will look like everything went well) Feb 08 19:38:20 Borromini: good idea, thank you, I'll regenerate images now with | tee - build.log and search for any size problems Feb 08 19:43:31 xback: and as a result you get the same tx queue timeout warning on ar71xx as well ? Feb 08 21:23:02 valku: you can also just turn on build logs... you get a logs/ directory in the buildroot Feb 08 21:25:37 karlp: yes, I am aware of BUILD_LOG... I think what happened here is that I forgot that sysupgrade.bin images are not generated for bcm53xx target. I have a few other targets and I think I might have gotten confused Feb 08 22:17:52 ynezz: et al, summary of my openwrt rebuild times: http://palmtree.beeroclock.net/~karlp/bugs/owrt-rebuilds/ Feb 08 22:18:50 blogic: karlp: 43k lines? oh, it seems we have a SoftMAC hw & driver for 802.11ax! Feb 08 22:18:55 wait... Feb 08 22:26:03 Iv'e got full build logs for this stuff and can try anyones suggestions, but I'm not really sure what else to look at. Feb 08 22:37:24 karlp: I would probably do `make V=sc -j1 2>&1 | tee native-build.log` and diff -u native/vm build logs (2nd run) Feb 08 22:40:26 * karlp retries with sc, just s wsn't showing any diffs Feb 08 22:41:45 rmilecki: ath11k looks more like fullmac, I found 66 times the string "firmware" ;-) Feb 08 22:42:09 I think they were being sarcastic thaht with that much code, it should have been enough for softmac :) Feb 08 22:42:12 yeah, I know :) Feb 08 22:42:26 karlp: :) Feb 08 22:42:30 ok Feb 08 22:42:50 What is the difference between V=s and V=sc? It has never been clear to me Feb 08 22:43:03 c has way more shell cruft in the display Feb 08 22:43:39 oh.... cat drivers/net/wireless/broadcom/brcm80211/brcmfmac/*.[ch] | wc -l Feb 08 22:43:40 34557 Feb 08 22:44:14 ath11k will probably grow ;-) Feb 08 22:44:53 yeah, that's interesting... cat drivers/net/wireless/mediatek/mt76/*.[ch] | wc -l Feb 08 22:44:54 11330 Feb 08 22:45:19 ipq807x is a quite interesting SOC, unfortunately still with NSS/ NPU offloading Feb 08 22:46:21 ynezz: https://hastebin.com/idakibageh.rb for instance. there's no useful difference in logs, it just.... doesn't do something for some reason. Feb 08 22:46:31 both vm and native are make 4.2.1. Feb 08 22:54:23 karlp: make package/firmware/linux-firmware/{clean,compile} V=sc -j1 | tee build.log; pastebinit build.log -> http://paste.ubuntu.com/p/QPzsTvgqnz/ Feb 08 22:55:25 that's the log for diff I've meant Feb 08 23:17:07 yeah, but even just recompiling it, doing "nothing" is taking longer Feb 08 23:17:35 but which step is taking that 30s ? Feb 08 23:17:43 What does V=sc provide over V=s? Never been very clear to me Feb 08 23:18:02 c is more verbose output for cmake/kernel Feb 08 23:18:08 Ah, thanks! Feb 08 23:19:26 karlp: I've forget to add stderr redirection 2>&1, so it should've been longer `make package/firmware/linux-firmware/{clean,compile} V=sc -j1 2>&1 | tee build.log; pastebinit build.log` -> http://paste.ubuntu.com/p/4nXsSG4XFz/ Feb 08 23:19:54 _surely_ that's what's going into the build logs anyway right? Feb 08 23:20:00 I mean sure, I can do yours too :) Feb 08 23:20:26 if you take a look at the output, it's quite different then yours Feb 08 23:20:49 yours was clean,compile, not just compile, compile... Feb 08 23:22:27 ok, I see what you mean, try it with just compile Feb 08 23:24:50 only difference is the times. https://paste.ubuntu.com/p/C5k8jFKBbB/ Feb 08 23:24:55 which is why it never makes sense for anyone Feb 08 23:26:05 but the clean step takes ages as well Feb 09 00:24:33 no idea why though. Feb 09 00:25:18 * karlp wonders if tracker is perhaps doing odd things Feb 09 00:26:48 hrm, shouldn't, tracker claims to ignore directories with .git dirs Feb 09 01:36:57 OK, huge kudos to whoever ported dtc as a package! Feb 09 01:37:39 Anyone know how to build an OpenWrt package as a static binary, given that my target has an unknown libc on it? Feb 09 01:41:05 Will just adding -static to CFLAGS do it? Feb 09 01:42:41 (I'm trying to run dtc natively under OEM firmware, rather than hoping that tar-ing up the device tree and hoping that works.) Feb 09 01:44:29 do yourself a favour and just tar it up, you can easily copy it to a USB stick Feb 09 01:45:11 OK, never tried to run dtc on a copy of /proc/device-tree, but worth a shot! Feb 09 01:45:29 or if you don't have USB ports, use base64 encoding and copy'n'paste it from serial/ telnet Feb 09 01:45:32 decompiling a compiled device tree is very not-preferable Feb 09 01:45:42 it's a mess Feb 09 01:45:49 but it might still give some clues Feb 09 01:45:58 Pleasantly surprised today to somehow get a root prompt on my EA8300 Feb 09 01:46:28 Agree, not the way to *generate* a DTS, but provides "ground truth" as to what works on the device Feb 09 01:47:57 I had a (relatively short) attempt at trying to get into an old Samsung SMT-G3010 IAD, unfortunately the buggers put getty on the serial console, with an unknown root password (didn't spend much time on it and just put it into the recycling bin, AMAZON ME isn't supported anyways - and that ancient software wasn't doing anything sensible either) **** ENDING LOGGING AT Sat Feb 09 02:59:57 2019