**** BEGIN LOGGING AT Tue Mar 05 02:59:57 2019 Mar 05 03:02:30 mangix: pong. The forced 1G link is the one connecting GMAC1 to port0 of builtin switch. Mar 05 03:04:45 the second GMAC is connected to the switch? Mar 05 03:05:15 mangix: correct. Mar 05 03:05:25 so it's like mt7621 Mar 05 03:06:29 hrm the schematic I have indicates it's separate. but still doesn't explain why the ethernet is not working. Mar 05 03:07:08 oooohhh i see. nvm Mar 05 03:28:48 bah still not working Mar 05 03:29:16 gch981213: /proc/cpuinfo shows the device as 9533. bug? this was also the case in ar71xx Mar 05 03:29:50 the device is QCA9531 Mar 05 03:33:10 mangix: I guess QCA9531 and QCA9533 have exactly the same chip id and the first person submitting it named that chip QCA9533. Mar 05 03:34:26 mangix: Would you mind to paste your dts somewhere? Mar 05 03:35:09 gch981213: https://github.com/neheb/source/commit/5b9bce823823786b2e93f0b118fa6b16e6851f36 Mar 05 03:35:50 everything works except the ethernet Mar 05 03:36:06 i tried adding an eth0 entry, but that wasn Mar 05 03:36:16 't generating any interrupts. Mar 05 03:40:37 mangix: Is that a single-port router? If it is, which port is the one available? Mar 05 03:40:57 it's single Mar 05 03:42:16 FWIW I have the original ar71xx patch Mar 05 03:48:24 mangix: If 'swconfig dev switch0 show' shows only port0 is up even if you connect the ethernet port, you'll need to change eth nodes like this: https://paste.ubuntu.com/p/2gtHf6drRb/ Mar 05 03:53:55 gch981213: link: port:1 link:up speed:100baseT full-duplex auto Mar 05 03:54:06 it shows port 2 3 and 4 with link down Mar 05 03:55:23 mangix: could you show me the ar71xx patch for it? Mar 05 03:59:13 sent Mar 05 04:03:08 huh. just noticed port 0 has txflow and rxflow whereas port1 does not Mar 05 04:04:35 mangix: Change phy-swap and phy-addr-swap to 1 in the dts I pasted a moment ago and try if it works. Mar 05 04:27:22 gch981213: both did not work. in the 0 case, swconfig shows MB counters but the device shows no interrupts. in the 1 case, nothing gets connected. I have a feeling the hardware is just faulty. Mar 05 04:37:42 mangix: eth0 isn't link up when you set both bits to 1? I don't remember whether switch-phy-addr-swap works. If it doesn't, you'll need to set phy-handle to swphy0. Mar 05 04:44:54 ah QCA953X_ETH_CFG_SW_PHY_SWAP is not implemented in ag71xx Mar 05 04:46:51 mangix: The ar933x one is the one actually used. Mar 05 04:52:25 got it working Mar 05 04:52:42 it was swphy0 Mar 05 04:54:08 i copied ar9344_comfast_cf-e120a-v3.dts Mar 05 04:55:40 oddly enough, swconfig dev switch0 show does not show link up anymore Mar 05 04:55:48 only for port 0 Mar 05 05:19:07 mangix: What the switch-phy-swap bit did is swapping PHY0 and PHY4. PHY0 is the port available on your router and after that swap, you have PHY0 connected to GMAC0 and PHY4 connected to builtin switch, so all ports on switch are down. Mar 05 05:31:03 so GMAC0 is not connected to the switch? Mar 05 06:02:25 mangix: Correct. GMAC0 is connected to phy0, phy4 or not connected at all. Mar 05 06:30:16 gch981213: do you know how the clock is set for QCA9531? I think OpenWrt thinks it's running at 500MHz whereas it actually runs at 560MHz. This is what I mean. https://paste.ubuntu.com/p/bgcXdTKyc3/ Mar 05 06:30:39 dmesg has the entry "CPU clock: 500.000 MHz" Mar 05 06:37:16 hrm. bootloader says 500MHz Mar 05 08:06:44 when removing a luci-app-*, and it still is visible in webUI, one needed to delete something to force a refresh? Mar 05 08:07:01 I don't recall and a quick search online didn't yield much Mar 05 08:07:04 anyone recalls? Mar 05 08:14:39 jow: after I uninstalled luci-app-nlbwmon (but Bandwidth Monitor menu still visible), upon clicking Bandwidth Monitor I get errors (obviously since its not installed anymore, how can I refresh luci to remove the entry without restarting the box?) Mar 05 08:14:58 hi all, can uboot overide using bootargs params from device tree? Mar 05 08:17:51 abenz: probably restarting uhttpd would help Mar 05 08:18:07 ynezz: I tried that, didnt help Mar 05 08:19:54 abenz: then there's some persistent cache which needs to be deleted (I don't know offhand where it's stored in /tmp) but you can disable it in /etc/config/luci (look for cache=1) and then restart uhttpd again Mar 05 08:21:36 probably `rm /tmp/luci*` could help as well Mar 05 08:26:00 abenz: browser cache? Mar 05 08:27:52 I will give those a go Mar 05 08:28:48 this did it: Mar 05 08:29:04 rm /tmp/luci-indexcache Mar 05 08:29:08 uhttpd restart Mar 05 08:29:45 cheers Mar 05 08:36:46 linkit has console=ttyS2,57600, i need override them by uboot, because old uboot has 57600 and dtb in sysupgrade image too. currently the only way is to set in new uboot baudrate to old one Mar 05 08:52:10 indy: it should if your u-boot supports fdt Mar 05 08:58:56 indy: On ramips target, u-boot doesn't have access to the dtb used by kernel. In order to use device tree on legacy u-boot, OpenWrt patched dtb into a specific place of the final kernel image. U-boot can't override kernel cmdline in this case. Mar 05 09:00:07 mangix: It's calculated according to several PLL registers. Mar 05 09:05:10 ynezz, gch981213 thanks, i'm playing with uboot master where landed support for mt7688 Mar 05 09:05:56 'lo Mar 05 09:06:12 ynezz, gch981213 so what all needs to be changed to kernel/uboot/dtb to have all cooperating? Mar 05 09:25:34 lynxis: I'm wondering how open is open in coverity terms :) I'm not able to see details for any of the defects Mar 05 11:05:14 jow: wb :) Hope you didn't got detoxed too much without internet ;-) Mar 05 11:05:33 jow: can I push this later on? is it ready? --> https://git.openwrt.org/?p=openwrt/staging/xback.git;a=commit;h=cb4acd4a5830b118c4c4fb88fbe6d0fccbbf2664 Mar 05 11:29:44 xback: thanks. yes, please. while you're at it can you also fold https://bugs.openwrt.org/index.php?do=details&task_id=2122#comment6057 into the commit? Mar 05 11:30:09 and add "Fixes: FS#2122" before the S-o-b Mar 05 11:30:31 jow: will do Mar 05 14:45:11 nbd: ping - How up would you be for providing some assistance to me in making some changes to act_connmark? Mar 05 14:49:24 what kind of changes? Mar 05 14:52:06 lol - a wise question - I have this harebrained idea of trying to preserve DSCP across WAN/Internet connections by storing/restoring them to/from ct->mark (and hence into skb->mark) Mar 05 14:54:14 and also (obviously) into the diffserv field. Mar 05 15:04:39 interesting Mar 05 15:16:34 nbd: you've got mail - be sat down when you open the patch.... it's my code and I don't know how to do what I want...but I know what I want to do :-) Mar 05 17:08:37 Any experience with U-Boot (OEM) deciding to re-write the environment after boot_part was changed using `fw_setenv` under OpenWrt? Mar 05 17:09:26 Seems to only happen for a 1 => 2 transition and not for a 2 => 1 transition, and only the first time that transition is made Mar 05 17:09:41 Perhaps the checksum is wrong?? Mar 05 17:10:25 (Linksys EA8300 "Dallas" -- IPQ4019) Mar 05 18:23:17 lynxis: looks like you still need to be a member of the project to see the actual issues (not just the number of issues) on coverity Mar 05 19:45:26 Hello, Mar 05 19:46:29 How I should upload a simple bugfix of a device? I want to modify the DTS to limit the range of freq supported by the device Mar 05 19:47:26 should I* Mar 05 19:56:17 vk496: https://openwrt.org/submitting-patches Mar 05 20:02:15 pkgadd: yes, already checked. But not really info about what to do with this kind of little bugfixes: Should mention the device in the title of the commit? the target (ramips in this case)? or a generic word like "bugfix:"¿ Mar 05 20:04:46 look at examples in the commit log or mailing list, e.g. [PATCH] ramips: fix blablubb initialization for the whateverCorpDevice Mar 05 20:06:16 pkgadd: okay, thnks! will check Mar 05 20:34:28 I noticed that there are a lot of devices with generic limitation of 5Ghz freq ("ieee80211-freq-limit = <5000000 6000000>;"), and probably is not true for most of them. Should this entry be fixed based on each device "Technical details" of the manufacter? Mar 05 20:49:19 Checking the ART/board file as well -- QCA9888 on Linksys EA8300 is limited to 5500 MHz and up Mar 05 20:50:34 Look at a62(?) in IPQ4019 as well -- it's limited in its band coverage in the DTS already Mar 05 21:08:59 jeffsf: you might need https://patchwork.ozlabs.org/patch/1051204/ too Mar 05 21:19:11 Thanks lach1012 Mar 05 21:20:40 I submitted https://patchwork.ozlabs.org/patch/1050950/ to pick a viable channel, as 36 (and VHT80) isn't appropriate in all cases Mar 05 21:58:22 vk496: the reason for ieee80211-freq-limit is usually to limit functionality of one radio to the 5 GHz (and the other to 2.4 GHz), because while the chipset itself could do both, the rf filters, amps, antennas, etc. are only tuned to a single band. yes, there are some examples where you have two 5 GHz radios tuned to different subsets of the band (e.g. netgear r8000) which warrant a more exact Mar 05 21:58:28 value, but in general 5000000-6000000 is fine (and mac80211's channel selection code does the finer grained stuff) Mar 05 22:31:05 i'm wanting to turn off ipv6 on just my lan interface, is the "ipv6" option for that interface in /etc/config/network the correct way to do this Mar 05 23:06:57 malwar3hun73r_: "Turn off" meaning (a) Don't offer IPv6 routing, (b) Don't accept any IPv6 traffic, (c) Disable Ipv6 completely, including the mandatory link-local IPv6 address, (d) something else? Mar 05 23:11:03 malwar3hun73r_: https://forum.openwrt.org/search?q=disable%20ipv6 Mar 05 23:16:45 jeffsf, well, honestly, i'm not sure i need it entirely and definitely don't need it on the LAN, do i? Mar 05 23:28:26 For most users of OpenWrt, IPv6 is "invisible" -- it either just works and is reasonably secure, or they don't have IPv6 with prefix delegation, so it doesn't do any harm Mar 05 23:29:25 I was pleasantly surprised when my IPS started offering IPv6 and "magically" my LAN clients were connecting over IPv6 Mar 05 23:29:32 s/IPS/ISP/ Mar 05 23:43:06 yep Mar 06 01:43:09 ^ different build keys on the builders signing some of the sysupgrade images? Mar 06 01:47:10 Any suggestions at tracking down why OEM U-Boot decides to spontaneously "Saving Environment to NAND..." ? Mar 06 01:47:23 Checksums on the MTD partition look good Mar 06 01:54:00 `strings` on the boot lodaer binary reveals `setenv boot_part %d;saveenv` Mar 06 01:59:57 (making the "offending" change in U-Boot triggers it, so not an OpenWrt-level thing) Mar 06 02:38:19 my ISP has been pretty bad lately and I've had to use network > interfaces > restart on the WAN interface in LuCI a lot. Is this equivalent to a `ifdown/ifup wan`? Mar 06 02:39:23 my bad, wrong channel! **** ENDING LOGGING AT Wed Mar 06 02:59:57 2019