**** BEGIN LOGGING AT Fri Feb 07 02:59:57 2020 Feb 07 08:57:53 jow: ping - collectd problem (?) Feb 07 08:59:21 ldir: pong Feb 07 09:01:06 noticed you've changed luci/collectd implementation. I updated last night, this morning all my graphs are 'NaN' - any ideas? Feb 07 09:01:27 collectd running :-) Feb 07 09:04:06 ldir: hmm, I can't see right now how the LuCI change would affect this Feb 07 09:04:53 ok, let me poke a bit more... I've probably done something silly Feb 07 09:14:21 KanjiMonster: Yet, the build error indicates it did .. switching to 5.4 fixed it. I didn't test 5.2. kernel would be 5.4 with backports from 5.4-rc8 + ath10k 5.4. seems the closest we've ever gotten :-) Feb 07 09:37:58 jow: doing a make clean & rebuild after clearing ccache - wondering if I've got some sort of frankenbuild after trying to build with 5.4 Feb 07 10:42:49 Aha! clue - /usr/bin/stat-genconfig isn't generating much output. Feb 07 10:44:26 ahhh! possibly light dawns... hold on Feb 07 10:45:24 ha! solved. Feb 07 10:46:05 ldir: what is the issue? Feb 07 10:46:13 local name = filename:gsub("%.lua", "") Feb 07 10:46:20 meh Feb 07 10:46:21 should be '.json' Feb 07 10:46:22 ok Feb 07 10:47:45 fixed Feb 07 10:47:51 he he Feb 07 11:07:07 thanks for the fixup :-) Feb 07 11:44:26 WARNING: Makefile 'package/network/services/hostapd/Makefile' has a dependency on 'PACKAGE_wpad-mesh-openssl', which does not exist Feb 07 13:26:46 I took a closer look at how the TP-Link Web UI updates the WA850RE (and thus inside the vendor provided fw update) and realized that it should use tplink-safeloader (instead of -v1) Feb 07 13:27:33 urjaman: That makes sense, and then tplink_hwid might actually be 0 Feb 07 13:28:22 Talking about it, it was strange anyway that they would have gone back from tplink-safeloader for v2 to tplink-v1 for v4 ... Feb 07 13:30:46 I only made the connection to safeloader after googling for "fwup-ptn" as in "has anyone decoded this partition table thing that contains this string" :P Feb 07 13:31:26 but that would also explain why I got a 0x0 hwid with binwalk Feb 07 13:32:25 yeah i had previously just analyzed my dump of the actual flash contents, not the fw update really Feb 07 13:33:40 oh a question on the kernel load addr / entry point Feb 07 13:34:08 i looked at the serial port while vendor fw started and it said "text base: 80000000" "entry point: 8000c150" Feb 07 13:39:18 bbl i need food (but in short, should we try to mimic that entry point? how does it work? i see no point to...) Feb 07 13:39:53 I cannot help you there Feb 07 13:41:48 but IIRC you do not need TPLINK_FLASHLAYOUT at all there, but have to add your setup to tplink_safeloader.c Feb 07 13:42:15 There are several tplink-safeloader devices in ramips, just have a look at how they do it. Feb 07 13:43:02 typically, the only variable you need in image/ is TPLINK_BOARD_ID, which references the correct config in tplink_safeloader.c Feb 07 13:45:33 yup Feb 07 13:46:46 is there some rules for the board id? like i see both TL-WRwhatever and TLWR being used etc Feb 07 13:47:31 also RE200-V1 vs TLWA850REV2 and so on Feb 07 13:48:45 i guess i'll go with TLWA850REV4 just so it sorts nicely among the others there Feb 07 13:49:03 I do not think there are rules there. This is from the times when nobody cared about such stuff. I'd say just do it like the closest relative. Feb 07 13:49:34 yes, that should be best. Feb 07 13:50:37 oooh - changes to /usr/share/fw3/helpers.conf don't survive a sysupgrade Feb 07 13:51:13 ldir: what changes were needed there? Feb 07 13:51:54 Maybe you don't want all the installed helpers enabled? Feb 07 13:52:37 you can explicitely configure the helpers that should be enabled for a zone Feb 07 13:53:24 oh can you? Feb 07 13:54:23 https://pastebin.com/vcB63ZsS Feb 07 13:54:30 to be fair I don't think I really understand them! Feb 07 13:56:35 and/or the security implications Feb 07 13:57:21 yeah, ftp can have security implications for example Feb 07 13:57:46 since it allows secondary data streams through the firewall if a control packet on the primary connection indicates so Feb 07 13:58:41 but you'd normally have the helpers enabled on the wan zone, not the lan zone? Feb 07 14:03:37 fw3 should enable them on all non-masq zones Feb 07 14:03:43 so only on internal ones Feb 07 14:04:39 yes, that's what it has done, for all the helpers in the 'lan' zone Feb 07 14:05:07 ah now I got your remark Feb 07 14:05:19 my paste example was wrong, it should be lan, not wan of course Feb 07 14:05:45 ah, ok :-) Feb 07 14:14:44 updated the wiki doc a little with the new found options. Feb 07 14:27:00 xback: Hauke: I did unbrick the UAP-LR, so I can test it with 5.4 again once those changes are merged. Feb 07 15:09:27 *sigh* Feb 07 15:09:59 I bricked the WA850RE with the web update thing... Feb 07 15:10:15 time to get electronical Feb 07 15:19:08 (and by bricked i mean it doesnt u-boot, so i need to do some in-system flashing...) Feb 07 15:22:14 if you provide your current patch/commit status, I can have a quick look Feb 07 15:23:50 i'll update the PR so you can take a look but add a warning to the commit message for now Feb 07 15:24:55 done Feb 07 15:34:42 {"default-config", 037e0000, 0x10000}, Feb 07 15:35:16 should by 0x3e0000 in the middle Feb 07 15:35:45 And there is a 0x2000 gap before partition-table Feb 07 15:36:25 ok the default config is a typo ... dunno why that'd break it to but i'll take a look Feb 07 15:37:02 and the gap is known but maybe i shouldnt have it in that fw (it's just it doesnt align with 64k blocks so that 8kb is useless for openwrt) Feb 07 15:37:12 *in that fw ... in that table .. Feb 07 15:38:06 i'll get back to trying to put in a real reset line to this thing Feb 07 15:49:22 adding a reset line and thus freedom to read the SPI with an external flasher achieved ... yeah it wrote ... something ... over parts of the u-boot (0x0020 to 0x10020, roughly) Feb 07 15:50:30 i'll try to just reset this thing to my original backup, fix that partition table, make a new binary and retry i guess Feb 07 15:50:35 then analyze more if it breaks again Feb 07 15:51:34 well, if the offset was wrong, it might be that it counted again from the beginning after it reached the end Feb 07 15:52:10 Though I don't know what "037e0000" will actually become. Feb 07 15:52:22 yeah i was thinking it might've tried to write that and wrapped around (i dunno why it would be trying to write the default config partition tho...) Feb 07 15:53:01 anyway, try the correct config and then we'll see Feb 07 15:54:56 i'm kinda impressed i was able to poke that reset line into the tiniest resistor i've observed so far :P (i had located it already previously because curious as to whether it was possible to have a real reset on this thing) Feb 07 15:55:17 the cheap digiscope i have helped but still Feb 07 16:02:25 yeah, those wa850 devices are annoying for a lot of reasons Feb 07 16:13:56 oh that's a worse typo than i thought at first look Feb 07 16:15:06 037e0000 means is like octal 37 exponent 0 ... which i think decodes to 31 but i'd need to check Feb 07 16:17:58 and what happens to the "e" then? Feb 07 16:20:47 That's actually interesting: What will c compiler do on a literal without 0x but with a letter in it? Feb 07 16:23:21 that's the exponent character in this case i think, but i didnt know one could have it in octal numbers Feb 07 16:23:50 and yeah my first thought was "how did an octal number with 'e' in it compile???" Feb 07 16:24:55 but yeah now with correct partition table, the upgrade from TP-link firmware to openwrt worked as-is :) Feb 07 16:25:03 just sent the -factory.bin to the web UI Feb 07 16:25:26 and it boots to openwrt afterwards, kinda neat Feb 07 16:27:56 Oh i have a question/suggestion related to the 4M flash devices with SPI flash but not to this one in particular ... Feb 07 16:29:10 given how JFFS2 needs it's empty erase blocks, wouldnt it make sense to set the kernel (for the small flash device builds) to use 32kB erase size where available instead of 64kB? Feb 07 16:30:22 it wouldnt be a big patch to have it pick the 32kB eraser from SFDP (it doesnt have the info about 32k erasers for flash chips without SFDP, but afaik most modern SPI chips have sfdp) Feb 07 16:33:21 urjaman: oh, you are right, it's the exponent. So every other letter would have caused an error ;-) Feb 07 16:34:25 well, octal 37 is hex 0x1f, so here you get your 0x20 Feb 07 16:34:57 yup Feb 07 16:35:38 and the TP-link firmware (on the serial port) said something about setting a dirty flag in the default config partition when it finished the "upgrade" Feb 07 16:35:45 so it all makes sense in the end Feb 07 16:36:26 for the erase blocks I can't help you, maybe write to the devel list or wait on somebody else's feedback here Feb 07 16:36:51 you can remove the 4Mmtk from the PR then Feb 07 16:37:11 oh yeah will remove the 4Mmtk Feb 07 17:40:02 xback: assuming you have a backports bump in your queue, wouldn't the old kernel also use the newer backports and need the newer branch? Feb 07 19:54:43 xback: I've a 'for-koen' branch in my staging - has a fix for building on macos, which helps, but another fix is needed. Getting closer. Feb 07 20:45:21 Hi any one know what's going on with this? Feb 07 20:45:23 https://github.com/openwrt/openwrt/pull/2756 Feb 07 20:45:38 I thought it was simple! Feb 07 20:45:41 lol Feb 07 20:45:46 Silly me! Feb 07 20:46:33 When I built it for my wrt3200acm it worked, but some people are saying there is somthing rong with the patch. Feb 07 21:23:27 my tree seems screwed up somehow. make defconfig fails Feb 07 21:26:23 http://sprunge.us/OSQmXb ... one of the errors seems to be in a package definition called PACKAGE_prometheus-node-exporter-lua-hostapd_stations Feb 07 21:29:38 ah, yeah, it's in packages feed Feb 07 21:40:42 russell--: maybe you need this: https://github.com/openwrt/openwrt/pull/2713 Feb 08 00:06:00 Pepe: yes, that looks likely. Just updated the build host to gnu make 4.3 **** ENDING LOGGING AT Sat Feb 08 02:59:58 2020