**** BEGIN LOGGING AT Sun Apr 04 03:00:04 2021 Apr 04 04:30:35 rsalvaterra: ping Apr 04 05:45:44 Is there a way to trigger a separate service to reload on a service_reload()? Apr 04 05:46:06 I would like to restart suricata once suricata-update finishes Apr 04 07:25:58 Build [#20](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/12/builds/20) of `oxnas/ox820` completed successfully. Apr 04 07:47:19 Build [#21](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/17/builds/21) of `ipq40xx/mikrotik` completed successfully. Apr 04 08:09:27 Build [#21](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/60/builds/21) of `armvirt/64` completed successfully. Apr 04 08:28:37 Build [#20](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/49/builds/20) of `bcm53xx/generic` completed successfully. Apr 04 08:38:07 Build [#20](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/46/builds/20) of `ath79/nand` completed successfully. Apr 04 09:41:02 Build [#20](https://buildbot.openwrt.org/openwrt-21.02/images/#builders/30/builds/20) of `x86/generic` completed successfully. Apr 04 10:18:14 xdarklight: I think we are not using the latest GPHY FW in OpenWrt, the user with the 7412 could try the most recent one Apr 04 10:34:49 guidosarducci: sorry can you add libelf as a tools package? it's not really installable on macs so we should have it in tools Apr 04 10:36:08 aparcar[m]: a 'libelf' is available on MacOS through the 'brew' package manager. Had some long discussion with Github user httpstrom around this. Apr 04 10:36:25 but libdw? Apr 04 10:36:29 i couldn't find that Apr 04 10:36:40 aparcar[m]: ^^^ httpstorm I mean Apr 04 10:39:17 aparcar[m]: there's a dwarf lib too I think, but BTF support, and then pahole building needs the elfutils versions of libdw. I don't think anyone can build elfutils on MacOS, so BTF is not really supported. Make things sort of moot, unless you know something? Apr 04 10:39:56 ynezz: ping? Apr 04 10:40:04 ynezz: ^^^ Apr 04 10:42:49 aparcar[m]: FYI see this comment and environs for some context: https://github.com/openwrt/openwrt/pull/3855#issuecomment-790232166 Apr 04 10:45:45 aparcar[m]: BTW the libdw check is guarded by a test on CONFIG_DWARVES, so not normally enabled (on MacOS). We could make doubly sure by disallowing building dwarves in Kconfig when on MacOS, but I can't see how to do that. Is it possible? Apr 04 10:47:01 aparcar[m]: IOW, can we have a build-host dependency in OWRT Kconfig? Apr 04 10:47:03 I'll wait for ynezz response, for me this all seems fine but jow suggested to build it as a host tool, in these cases I just stick to what the gurus suggest Apr 04 17:10:48 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Apr 04 18:02:07 nbd jow: what's the policy on what can be put into /lib/functions/ ? there's some shared parsing for UCI for the strongswan ipsec and swanctl CLI's that I'd like to factor into /lib/functions/strongswan.sh if that's admissable... Apr 04 21:19:47 Hauke: re 7412: he's saying that everything is fine with the old out-of-tree GSWIP/PMAC driver. if you have any hint where to get newer GPHY firmware from I can ask him to test that Apr 04 21:21:37 Hauke: re 7360: even after the 50MHz rate patch we seem to get MMDIO_STAT = 0xf7 and MAC_PSTAT = 0xd38 Apr 04 21:39:10 Hauke: so I am wondering if we should switch away from PHY auto polling and configuration onto manual configuration like the out-of-tree driver did - at least for testing on the 7360 Apr 04 21:47:59 xdarklight: yes expicitly configuring could work Apr 04 21:51:40 xdarklight: the PHY is detecetd as 1G capabale and the MAC is configured like this Apr 04 21:52:44 the AR8030 supports 1G Apr 04 21:54:43 Hauke: according to this marketing slide it shouldn't support 1G: https://content.codico.com/fileadmin/media/download/datasheets/qualcomm/qualcomm-ethos-AR8030.pdf Apr 04 21:54:59 Are you sure it is using RMII and not RGMII? Apr 04 21:55:23 ah ok, I saw a difefernt version Apr 04 21:56:33 Hauke: AFAIK all other PHYs from the AR803x series are Gbit/s PHY. the exception seems to be AR8030 Apr 04 21:57:32 Hauke: argh, I need to add GSWIP_MII_CFGp to the register dump so I am not 100% sure if the registers are programmed to RMII mode. according to the .dts they are in rmii mode though Apr 04 22:01:59 xdarklight: based on your register dump I think the MAC is right to assume 1GBit/s Apr 04 22:02:06 but this will not work with RMII Apr 04 22:02:30 Hauke: so precisely you mean "this will not work with RMII + auto-polling"? Apr 04 22:02:47 yes I do not think this will work Apr 04 22:03:35 We have to either add some code which detectes that autopoling wants to do 1G on RGMII and then select 100Mbit/s instead of completly configure this PHY manually Apr 04 22:03:57 xdarklight: where do I find your code for fixed link configuration? Apr 04 22:05:38 Hauke: still in rough shape: https://github.com/xdarklight/linux/commit/f2cf732dd2a583317114ae229f4b1622b3863c21 Apr 04 22:05:52 Hauke: I'll drop the if for fixed links and try it on my HH5A Apr 04 22:06:11 Hauke: also I think I need to combine it with the logic from gswip_phylink_mac_config Apr 04 22:26:18 xdarklight: Can we unset this BMSR_ESTATEN bit in the phy? Apr 04 22:29:20 Hauke: I'd expect it to be read-only. at least for most PHY datasheets I have seen BMSR is marked as RO (but I don't have a datasheet for the AR8030) Apr 04 22:31:04 yes read only Apr 04 22:31:06 too bad Apr 04 22:35:13 xdarklight: the ethtool output for this device would be interresting Apr 04 22:35:47 Hauke: which one precisely? Apr 04 22:39:12 the default page Apr 04 22:39:29 with the link modes Apr 04 23:12:32 Hauke: https://brpaste.xyz/LxUDkg/raw Apr 04 23:33:05 why does Linus only show 10 and 100 MBit/s? Apr 04 23:33:09 *Linux Apr 04 23:39:45 any idea how to debug "OF: /palmbus@300000/gpio@600: could not find phandle" ? Seems to only have started with 5.10, but looks like a DTS issue? Apr 04 23:40:53 Hauke: I am assuming that internally it does not just care about the PHY modes but also about the data from gswip_phylink_validate where we mask out the Gbit/s speeds for RMII PHYs Apr 04 23:41:18 xdarklight: ok makes sense Apr 04 23:41:45 the most recent PHY FW files are in the ltq_fw_PHY_IP-2.2.tar.gz file Apr 04 23:42:01 https://sources.openwrt.org/ltq_fw_PHY_IP-2.2.tar.gz Apr 04 23:42:33 lipnitsk: Yes, that's a DTS issue. The kernel brings a lot of DTS files. Apr 04 23:43:36 Thermi: thing is I didn't touch DTS... but something may have gone out of date - this is an ancient RT2880 device (Belkin F5D8235-4 v1) Apr 04 23:44:09 lipnitsk: evidently for some reason the DTS changed. So look at which DTS is used by your box and track it down Apr 04 23:44:29 the dts file defines a ton of GPIOs and they all look the same, yet two of these errors appear at start... Apr 04 23:44:38 Come on, it specifies 600 Apr 04 23:44:55 copy it over to your box and use a text editor to find the structure Apr 04 23:45:01 you need to go through the includes it uses too Apr 04 23:45:05 * lipnitsk < https://matrix.org/_matrix/media/r0/download/matrix.org/ljwFPMwkAMbkxECmZlZygYrh/message.txt > Apr 04 23:45:44 if you have the dtc binary, you can decompile the dtb it uses to a dts file and look at that with a text editor. Apr 04 23:45:59 It contains everything pulled in via the includes at compile time Apr 04 23:46:10 and the decompiled dtb file is quite readable Apr 04 23:46:49 looks just like the others... I'll dig more. Apr 04 23:47:17 Thermi: thanks for the advice. will try that Apr 04 23:47:37 It won't get fixed by complaining :P Apr 04 23:48:06 Thermi: oh I know. I only complain (or ask for advice from someone with more expertise) after spending a few quality hours myself Apr 04 23:48:27 Thermi: appreciate the response ;) Apr 04 23:48:46 I only know the work flow because I had troubles with DTS files at work already ;) Apr 04 23:51:41 Thermi: yeah, the DTC compiler output is already helping ;) See, that's exactly the kind of help I needed Apr 04 23:51:53 thanks Apr 04 23:51:57 yw Apr 05 00:18:46 Hauke: https://github.com/xdarklight/linux/commits/lantiq-random-network-fixes-20210404 - see especially: https://github.com/xdarklight/linux/commit/4a978d25e17b6ee4f920142a75531ab7531f83b2 Apr 05 00:19:06 Hauke: testing on my HH5A shows: LAN1 (GPHY) and LAN4 (RGMII PHY) are working at 1Gbit/s, 100Mbit/s and 10Mbit/s Apr 05 00:19:24 Hauke: let me know if you see any issues, else I'll let the 7360 user test this (slightly modified for Linux 5.4 though) Apr 05 00:22:43 Hauke: uh, found a bug myself... Apr 05 02:20:02 The smell, feel, speed of a freshly wiped computer **** ENDING LOGGING AT Mon Apr 05 02:59:56 2021