**** BEGIN LOGGING AT Thu Sep 27 02:59:59 2018 Sep 27 03:31:16 I'm no getting a lot of messages like: Sep 27 03:31:17 ath10k_ahb a000000.wifi: rate-override: Skipping un-supported device-id, hw-nss: 2 dev-id: 0x0 Sep 27 03:37:12 They look like the following messages from QCA's firmware: Sep 27 03:37:12 ath10k_ahb a000000.wifi: Invalid VHT mcs 15 peer stats Sep 27 03:38:46 But they seems to appear less often than with the QCA firmware Sep 27 03:46:17 luaraneda: context? Sep 27 03:51:28 mangix: Testing the mac80211 update, and the replacement of (ath10k,QCA firmware) by (ath10k-ct, CT firmware) Sep 27 03:55:52 I'm updating to the latest changes and repeating the tests, but my phone seems to be constantly disconnected with the following message: Sep 27 03:55:53 daemon.info hostapd: wlan0: STA [MAC_ADDR] IEEE 802.11: disconnected due to excessive missing ACKs Sep 27 03:56:23 I'm not sure if it happened before, so I'll make some test with older builds too Sep 27 03:57:40 Ah ok Sep 27 05:50:07 how i can compile fwtool for host? Sep 27 05:55:16 make -j1 V=s package/system/fwtool/compile wants .config to be exist Sep 27 06:33:51 ds_shadof: add DEVELOPER=1 Sep 27 06:35:58 blogic:ping Sep 27 06:40:06 dedeckeh: hi Sep 27 06:50:53 blogic, tried DEVELOPER=1 make -j1 V=s package/system/fwtool/compile and make DEVELOPER=1 -j1 V=s package/system/fwtool/compile doesn't help Sep 27 06:53:26 why dont you just create a .config file ? Sep 27 06:57:37 blogic, i just need to compile some host binaries for my project. Is there any automated way to generate .config. I don't care about platform and targets Sep 27 06:59:26 make menuconfig; save; exit Sep 27 06:59:38 or maybe just make defconfig Sep 27 07:03:49 cp target/linux/generic/config-4.14 .config; echo CONFIG_TARGET_ar71xx=y > .config; make defconfig Sep 27 07:04:23 that first step is completely wrong and useless too since you overwrite it in the next Sep 27 07:05:53 hm, yes, so i don't even need to copy at first step since it works :) Sep 27 07:27:27 sooo Sep 27 07:27:36 what was changed in the build code recently Sep 27 07:27:43 coz it breaks kernel builds upon errors Sep 27 07:28:15 i run make V=99, kernel build breaks due to error, i fix error run make V=99 and it does not try to rebuild the kernel Sep 27 07:28:35 i need to manually call target/install which causes a full kernel rebuild Sep 27 07:32:08 stintel, o/ Sep 27 07:32:34 nitroshift: \o Sep 27 08:15:55 ds_shadof: the first config is a kernel config, not the openwrt one, luckily you overwrote it ;-) Sep 27 08:16:20 blogic: as far as I know nothing there should have been changed Sep 27 08:51:25 someone just offered to give me an obsolete ubnt uap-ac, apparently with broadcom radios. Sep 27 08:53:36 from the photos, it looks like the radios are in mini-pcie sockets, so, maybe replaceable. Sep 27 09:16:26 how do i debug procd doing nothing? /etc/init.d/mything start does nothing. no log output Sep 27 09:18:34 INIT_TRACE=1 /etc/init.d/mything $action Sep 27 09:18:38 thanks! Sep 27 09:18:42 iirc Sep 27 09:18:57 jup works Sep 27 09:23:13 wow, that's heaps more than I new about from PROCD_DEBUG=1. Sep 27 09:23:21 * karlp has added it to the procd init scripts page. Sep 27 09:23:25 karlp: thanks Sep 27 09:23:36 karlp: that has been my #1 issue with procd since its inception Sep 27 09:23:47 near impossible to debug Sep 27 09:23:57 and then I accidentally found INIT_TRACE somewhere in some file Sep 27 09:24:16 this is.... just adding echo to every bit of shell. Sep 27 09:24:27 I mean, that helps sometimes but it's not very good as far as debug goes. Sep 27 09:24:32 still, more tricks to use. Sep 27 09:27:42 package/base-files/files/etc/rc.common:${INIT_TRACE:+set -x} Sep 27 11:31:09 Hi all, having issues with yaffs2 filling up on old routers due to meta data. The issue is on upgrade the mtd partition is not erased and remounted but instead the root directory is simply overwritten by untaring the contents. Is it possible to unmount / when doing an sysupgrade? and if so could you please point me to examples. Thanks Sep 27 13:04:49 Neighbor11111113: OpenWrt doesn't use yaffs2 anymore since a while, maybe try a more recent release Sep 27 13:05:53 thanks KanjiMonster - I'm stuck with yaffs for now unfortunately. But looking more into the issue I think I may have misunderstood something Sep 27 13:07:13 jow:ping Sep 27 13:45:56 greearb: ping Sep 27 13:56:21 huaracheguarache: ping Sep 27 13:56:30 ldir: pong Sep 27 13:57:04 saw your question the other day about cake and ipv6 6rd Sep 27 13:58:22 I don't have definitive answer but there is/was a bug in the kernel to do with decapsulating ipv6 packets that can upset qdiscs associated with an ifb.... ie. cake Sep 27 13:59:07 ok. I guess the best way to find out is just to give it another try. Sep 27 13:59:11 there's a fix gone into upstream kernel, that should trickle down to us, but also a workaround fix has gone into the cake kmod, which can be removed when the kernel fix arrives. Sep 27 13:59:27 aha Sep 27 13:59:53 the symptoms from what I understand are 'slow traffic for ipv6' whilst ipv4 is ok. Sep 27 14:00:36 I didn't really try to differentiate between ipv4 and ipv6 traffic. the only thing I noticed was that browsing the web was pretty much impossible Sep 27 14:00:38 I know it affected 6in4 - I don't know if 6rd was affected but I don't really see why it wouldn't. Sep 27 14:01:37 if you look at the cake bumps that I put into master & 18.06 you'll see the clues. Sep 27 14:02:05 https://github.com/dtaht/sch_cake/commit/42e87f12ea5c390bf5eeb658c942bc810046160a Sep 27 14:02:26 ooo, nice Sep 27 14:03:00 https://marc.info/?l=linux-netdev&m=153684989427100&w=2 Sep 27 14:42:32 thanks =) Sep 27 15:18:30 huaracheguarache, here Sep 27 15:18:49 greearb: hi Sep 27 15:19:28 so I was roaming a bit around in the download directory on the candela website and came across http://candelatech.com/downloads/ath10k-9984-10-4/ath10k-fw-beta/firmware-5-ct-full-htt-mgt-community.bin Sep 27 15:19:56 I thought I would give it a try since it's a newer release than the beta firmware I'm running Sep 27 15:20:40 In the release notes it says that it has the mfp flag set, but it doesn't show up for me: Sep 27 15:20:46 ath10k_pci 0000:01:00.0: firmware ver 10.4b-ct-9984-fH-011-f9581cd api 5 features peer-flow-ctrl,txstatus-noack,wmi-10.x-CT,ratemask-CT,regdump-CT,txrate-CT,flush-all-CT,pingpong-CT,ch-regs-CT,nop-CT,htt-mgt-CT,set-special-CT,cust-stats-CT crc32 c019071e Sep 27 15:21:15 and speeds are low Sep 27 15:22:11 the '10.4b' firmware should not be in that dir Sep 27 15:22:58 I'll do full builds of the beta firmware soon-ish and will double-check the mfp flag is set. Sep 27 15:25:33 what's the difference between ath10k-9984-10-4/ and ath10k-9984-10-4b/? they both contain ath10k-fw-beta/ Sep 27 15:25:48 4b is a rebase of my patches on newer upstream firmware Sep 27 15:26:01 easier to bisect problems, but still needs more testing Sep 27 15:26:20 mkresin, ping (jow said the issue is your turf) Sep 27 15:26:50 ok. ping me if you want me to test anything Sep 27 19:43:29 greearb: hi Sep 27 19:44:06 greearb: I got a warning about "invalid rate bw=2, mcs=0, nss=1" with todays OpenWrt master and I think this is related to ath10k-ct: https://pastebin.com/9yRn0FQh Sep 27 19:44:53 it could also be related to ath9k, iwinfo wlan0 assoclist shows my Intel 6300 with 5Mhz TX channel which is strange Sep 27 19:45:15 wlan0 is using ath10k-ct Sep 27 20:13:48 I'm getting this warning in the current snapshot r8165-3fa7e62 with ipq40xx/generic: Sep 27 20:13:53 https://pastebin.com/raw/LWKLAiUh Sep 27 20:24:49 Could this be related to https://cdn.kernel.org/pub/linux/kernel/v4.x/ChangeLog-4.14.72 2b7844ed3c7a6939090260eec43309035ed44a57 Sep 27 20:27:04 nnet: I got the same warning, OpenWrt switched yesterday from ath10k and mac80211 from kernel around 4.14 to mac80211 from kernel 4.19-rc5 and to ath10k-ct from 4.16 this is probably a new problem introduced by that Sep 27 20:33:00 @Hauke noted Sep 27 20:33:17 The gl-b1300 is not listed in this commit: Sep 27 20:33:18 https://github.com/openwrt/openwrt/commit/61b5b4971e7d6aab6b0695997c3b5aaf73c53b5f#diff-56084c5e657badcdff6ce29181444b61 Sep 27 20:33:42 But it was made the default: Sep 27 20:33:42 https://github.com/openwrt/openwrt/commit/61b5b4971e7d6aab6b0695997c3b5aaf73c53b5f#diff-56084c5e657badcdff6ce29181444b61 Sep 27 20:34:13 Effective only yesterday ... : Sep 27 20:34:13 https://github.com/openwrt/openwrt/commit/3fa7e62cec16393d09a914e3c36a14df8353c608#diff-56084c5e657badcdff6ce29181444b61 Sep 27 20:36:26 I'm not quite sure how to debug the problem as WDS-AP (4addr) effectively, it's working fine in r8137-e9d92bf1e1/ ath10k && ath9k, it appears not to be working in r8155-c662299bf9/ ath9k (the WDS-client can use either version) Sep 27 21:05:34 pkgadd: I think it is not the first time 4addr mode is broken after we updated mac80211, nbd could you please have a look at this Sep 27 21:51:09 yep, 4addr has been acting up a few times in the past Sep 27 21:51:44 still working on an easier to debug test setup with bt hh5a and tl-wdr4300 (2.4 GHz, to get ath10k out of the picture) Sep 27 21:57:08 I'm now seeing the warning too on an ASUS RT-AC58U Sep 27 21:57:49 I'm going to try different combinations of ath10k(-ct) + firmware(-ct) Sep 27 21:59:37 Ah, and calibration data (board-2.bin) too Sep 27 21:59:54 I removed and installed again ath10k-firmware-qca4019-ct and I'm not seeing these anymore on the three meshed 3 APs. Sep 27 22:03:42 When doing ping, I'm seeing ping spikes to 3000 ms to my phone, and nearly 40% packet loss Sep 27 22:04:03 nnet: you can confirm that easily by running md5sum over /lib/firmware/ and /rom/lib/firmware/ - if they're the same, your reinstalling of ath10k-firmware-qca4019-ct was only a placebo effect (it should be), if they differ, there'd be a problem in the packaging/ buildsystem Sep 27 22:06:42 8 Sep 27 22:21:16 pkgadd: they are the same, might it have been my own earlier uploaded firmware in /lib/firmware/ that survived the upgrade? Sep 27 22:33:49 nnet: everything but /etc/ is overwritten in a normal sysupgrade (any variant of it), unless you explicitly add it to /etc/sysupgrade.conf for safe keeping Sep 27 22:35:29 I may have pushed the firmware there along with config. Sep 27 22:44:38 I just manually replaced the firmware-5.bin (ct) with an older version (qca) and the ping problems seems to be resolved Sep 27 22:45:02 greearb: ^^ Sep 27 22:47:53 But the kernel warning remains Sep 27 23:04:06 Anyone know what these are with SNAPSHOT r8165-3fa7e62 on ipq40xx/generic: Sep 27 23:04:12 ath10k_ahb a800000.wifi: rate-override: Skipping un-supported device-id, hw-nss: 2 dev-id: 0x0 Sep 27 23:10:50 Anyone a fan of TARPIT? Sep 27 23:12:27 Yes, we can drop packets to ports we won't forward… but what if we had millions of routers tarpitting probes? that would really slow down botnets and hackers, wouldn't it? Sep 27 23:20:07 Ok, with the combination kmod-ath10k + ath10k-firmware-qca4019 everything seems to be working fine Sep 27 23:20:16 no kernel warning (yet), no disassociation and 0% packet loss on my phone Sep 27 23:20:37 + the printer/scanner is working again :) Sep 27 23:21:15 Hauke: greearb: ^^ Sep 27 23:21:53 At least for QCA4019, it seems that the combination kmod-ath10k-ct + ath10k-firmware-qca4019-ct is causing problems Sep 27 23:26:10 Wait, looks like the 5GHz radio is having problems with kmod-ath10k + ath10k-firmware-qca4019 Sep 27 23:26:12 :( Sep 27 23:27:47 iOS devices are disconnecting right after connecting with SNAPSHOT r8165-3fa7e62 on ipq40xx/generic (kmod-ath10k-ct + ath10k-firmware-qca4019-ct). Sep 27 23:39:35 luaraneda: check the board-2.bin, by using the older linux-firmware versions, the upstream committed board files for -in particular ipq4019- have gone missing Sep 27 23:42:28 Same as ath10k-firmware 48e1f777217a32a0bf7f9c6fd4df294e board-2.bin Sep 27 23:44:36 pkgadd: That might be it: Sep 27 23:44:37 - old build: 48e1f777217a32a0bf7f9c6fd4df294e board-2.bin Sep 27 23:44:37 - new build: abdbc086569b87cec97a440860b18c1f board-2.bin Sep 27 23:45:01 I'm gonna run one last test (I had already reverted to the older build) Sep 27 23:52:22 pkgadd: Yes! that did the trick. 5GHz is working again. Thanks for the idea :D Sep 28 00:39:04 AFAIKT the gl-b1300 wasn't removed from board-2.bin Sep 28 00:39:37 iOS devices are disconnecting immediately from SNAPSHOT r8165-3fa7e62 on ipq40xx/generic gl-b1300 Sep 28 00:39:59 The pervious snapshot was working fine. Sep 28 00:40:20 nnet: that assumption would be wrong Sep 28 00:41:31 I see no evidence in the commit logs that any devices were removed since this commit: Sep 28 00:41:52 https://github.com/kvalo/ath10k-firmware/commit/3af9ae584691acbe2356e30fac8380ad4840d8a6#diff-2fc6c6fb9ab5bf0381bb93c90cd84a47 Sep 28 00:42:27 you're 'just' missing the poing that ath10k-firmware no longer produces any binaries Sep 28 00:42:31 nnet: devices weren't removed, but the board-2.bin changed Sep 28 00:42:57 the binaries -and with that board-2.bin- are now provided by the linux-firmware source Sep 28 00:43:14 and that one is a lot older - lacking almost all recent ipq4019 boardfile additions Sep 28 00:43:22 and those are crucial for ipq4019 devices Sep 28 00:43:49 - ath10k-firmware-qca4019: abdbc086569b87cec97a440860b18c1f board-2.bin Sep 28 00:43:50 - ath10k-firmware-qca4019-ct: 48e1f777217a32a0bf7f9c6fd4df294e board-2.bin Sep 28 00:44:10 Exactly was pkgadd said Sep 28 00:44:41 Are you using only board-2.bin abdbc086569b87cec97a440860b18c1f from ath10k-firmware-qca4019? Sep 28 00:45:09 you need to use it from an *old* ath10k-firmware-qca4019 (or directly from kvalo's repo) Sep 28 00:45:37 because the new ath10k-firmware-qca4019 is provided by linux-firmware, with rather old board-2.bin Sep 28 00:45:52 I selected ath10k-firmware-qca4019 in menuconfig, but copied board-2.bin from ath10k-firmware-qca4019 -ct (48e1f777217a32a0bf7f9c6fd4df294e) Sep 28 00:46:04 kvalo's repo contains 48e1f777217a32a0bf7f9c6fd4df294e Sep 28 00:46:45 48e1f777217a32a0bf7f9c6fd4df294e /lib/firmware/ath10k/QCA4019/hw1.0/board-2.bin Sep 28 00:47:14 48e1f777217a32a0bf7f9c6fd4df294e ath10k-firmware/QCA4019/hw1.0/board-2.bin Sep 28 00:56:59 nnet: commit 9860cdd disabled kvalo's repo for QCA4019 an replaced it by linux-firmware repo, which is not updated frequently Sep 28 00:57:16 https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=9860cdda76797b0d9dc24795c39b307589589f4c Sep 28 00:58:23 luaraneda: I think what I was doing differently was trying to use still ath10k-firmware-qca4019-ct while you are using ath10k-firmware-qca4019 (with the latest board-2.bin). Sep 28 01:02:17 I'm now testing ath10k-firmware-qca4019 (with the latest board-2.bin) with kmod-ath10k-ct (while you switched to kmod-ath10k). Sep 28 01:02:19 It would be great if Kalle Valo could sync his repo with linux-firmare, but from what I've read, that might take some time Sep 28 01:03:04 nnet: I didn't test that combo, it might work Sep 28 01:03:24 I usually expect his repo is ahead of linux-firmware. Sep 28 01:03:45 luaraneda: iOS is happy now with that combo it seems. Sep 28 01:04:10 Ah, no. Sep 28 01:04:31 Now I'm getting those warnings again. :( Sep 28 01:04:59 Guess I'm going to have to try with kmod-ath10k, or ignore the warnings? Sep 28 01:05:01 The warning seems to be related to kmod-ath10k-ct Sep 28 01:07:00 I believe the warning can be ignored if your device is working correctly after the warning. If your network is unstable, you should try another combo Sep 28 01:07:49 It seems kmod-ath10k-ct should be used with ath10k-firmware-qca4019-ct. Sep 28 01:08:08 So I can test it a bit but eventually there like is some fix to be had. Sep 28 01:09:18 Yes, give it a couple of days, snapshots/master builds might break from time to time Sep 28 01:11:16 Yup. I'm curious as to why/if iOS 'dislikes' the -ct ath10k kmod+firmware. All other devices were fine with the correct board-2.bin. Sep 28 01:27:53 18.06.1 r7258-5eb055306f with ath10k/QCA4019/hw1.0/firmware-5.bin 10.4-3.5.3-00057 (over the 3.4 in ath10k-firmware-qca4019) shows the least warnings. Sep 28 01:37:31 luaraneda, you using .11w? I have not updated 4019 builds with latest firmware Sep 28 01:44:25 greearb: It was setting as optional (only a Windows client was able to connect before), but I disabled it and the problems continued Sep 28 01:44:50 Now it's disabled Sep 28 01:48:05 dmesg showed a lot of these messages (every time a device was disassociated/associated): Sep 28 01:48:06 rate-override: Skipping un-supported device-id, hw-nss: 2 dev-id: 0x0 Sep 28 01:50:18 and hostapd: Sep 28 01:50:19 daemon.info hostapd: wlan0: STA [MAC_ADDR] IEEE 802.11: disconnected due to excessive missing ACKs Sep 28 01:52:45 I first noticed it because Whatsapp web was losing the connection to my phone regularly Sep 28 01:54:58 Additionally, I couldn't connect to my printer, Epson L355 (WiFi) Sep 28 02:02:52 greearb: I don't know if it makes a difference, but I'm using wpad (full), instead of the default wpad-mini. Additionally, I have selected openssl Sep 28 02:27:23 I'm using wpad-mesh-openssl Sep 28 02:27:53 I'm seeing the rate-override messages. Sep 28 02:28:14 Critically, iOS devices specifically are auth'ing and immediately disconnecting. Sep 28 02:42:05 I'll build new images, see if that improves anything **** ENDING LOGGING AT Fri Sep 28 03:00:01 2018