**** BEGIN LOGGING AT Sun Oct 13 02:59:57 2019 **** BEGIN LOGGING AT Sun Oct 13 06:59:50 2019 Oct 13 08:06:15 I Oct 13 08:07:23 I'm running a mainline kernel on a BeagleBone Green and the Ethernet interface has issues where is doesn't come up all the time, reporting: Oct 13 08:07:26 [ 21.918837] net eth0: initializing cpsw version 1.12 (0) Oct 13 08:07:30 [ 21.926229] libphy: PHY 4a101000.mdio:00 not found Oct 13 08:07:32 [ 21.931261] net eth0: phy "4a101000.mdio:00" not found on slave 0, err -19 Oct 13 08:19:47 where would be the best place to report this bug? Oct 13 08:24:25 u-boot also reports: "Could not get PHY for cpsw: addr 0 Oct 13 08:24:26 " Oct 13 08:25:14 I noticed before that is: "Board: BeagleBone Black" Oct 13 08:25:21 which is not "Board: BeagleBone Green" Oct 13 08:51:59 rah: sounds like it's missing the (partial) workaround patch for the ethernet phy issue Oct 13 08:52:10 oh, "mainline kernel" .. yeah Oct 13 08:52:30 easy solution: use a beaglebone kernel :) Oct 13 08:53:14 I can understand the desire the run mainline, but some patches exist for good reason Oct 13 08:55:35 iirc spidev is also unusable on mainline, the sysfs gpio interface is substantially worse, gpios can't be setup in DT, pruss cannot be used (mainline uio-pruss doesn't support am335x, remoteproc-pru isn't in mainline yet either) Oct 13 08:55:57 there are probably many more reasons to use a beaglebone-specific kernel Oct 13 09:12:34 <@zmatt> I can understand the desire the run mainline, but some patches exist for good reason Oct 13 09:12:51 zmatt: such as? Oct 13 09:13:08 zmatt: what is the (partial) workaround patch for the ethernet phy issue? Oct 13 09:13:11 I literally just gave a list of examples, and you've also run into one issue yourself Oct 13 09:13:31 well obviously patches exist to make the thing work :-) Oct 13 09:13:49 I mean, why are the patches external patches and not applied to mainline? Oct 13 09:14:21 why do they exist as patches instead of commits in Linus' repository? Oct 13 09:14:25 because noone has had the time, energy, and motivation to try to go through the process of getting them into mainline Oct 13 09:14:50 I see Oct 13 09:15:07 in some cases the patches may be hacky and may need deeper thought to get them mainline acceptable Oct 13 09:15:26 e.g. if it has detrimental effects on other devices then that wouldn't matter for a beaglebone kernel, but it would be unacceptable for mainline Oct 13 09:15:46 zmatt: could you provide more specific information about "the (partial) workaround patch for the ethernet phy issue" so that I can possibly apply the patch and get the Ethernet interface working? Oct 13 09:15:50 (I don't know if any of the patches have detrimental effects on other devices, I'm just giving an example) Oct 13 09:16:11 yes, I understand about mainlining, thanks Oct 13 09:16:30 if you're going to run a patched kernel anyway, what's the reason to not just use rcn's -bone kernels? Oct 13 09:16:30 * rah has patches in the kernel himself Oct 13 09:17:38 uhh, check the patches list in the bb-kernel repo? it can't be hard to find Oct 13 09:17:42 I'd rather minimise the delta between the kernel I run and mainline Oct 13 09:19:15 $ ls patches/**/*phy* Oct 13 09:19:15 patches/drivers/ti/cpsw/0001-cpsw-search-for-phy.patch Oct 13 09:19:23 it was indeed not hard to find :P Oct 13 09:22:06 -rw-r--r-- 1 rah rah 3868 Apr 3 2016 ../debian/kernel/cpsw_search_for_phy.patch Oct 13 09:22:10 but most beaglebone users are not running mainline, so you're staying close to a branch of the linux kernel that's far less tested than if you use a -bone or -ti kernel, which is nicely demonstrated by the fact you've run into the ethernet phy issue Oct 13 09:22:12 it seems I've been here before Oct 13 09:22:29 doesn't TI pay people to upstream patches? Oct 13 09:22:43 TI does quite a bit of mainlining effort yes Oct 13 09:22:49 but the beaglebone is not a TI product Oct 13 09:23:04 (and this bug is beaglebone-specific) Oct 13 09:23:29 I'm saddened that this patch hasn't been upstreamed after three years Oct 13 09:24:10 so rectify it :) Oct 13 09:24:53 the hack done by the patch is conditional on of_machine_is_compatible("ti,am335x-bone") so it shouldn't negatively affect other devices Oct 13 09:25:31 so maybe mainline will affect it (after fixing obvious formatting issues) Oct 13 09:25:51 may want to cut out the CONFIG_OF_OVERLAY stuff, which never reached mainline and is even deprecated in beaglebone kernels Oct 13 09:27:25 *maybe mainline will accept it Oct 13 09:27:31 not affect :P Oct 13 09:27:57 I doubt it; hacking up the device tree at runtime doesn't seem like the kind of thing mainline would accept Oct 13 09:28:15 the kernel does it in a bunch of places already Oct 13 09:28:22 in mainline Oct 13 09:28:59 perhaps they would accept it but I would expect push-back Oct 13 09:29:12 if patching the DT were not acceptable, of_update_property() would not exist Oct 13 09:29:49 though now that I think of it, it would be better to patch the DT in u-boot probably Oct 13 09:30:15 u-boot ought to deal with this anyway to ensure netbooting isn't affected by this problem Oct 13 09:30:29 actually never mind that Oct 13 09:30:57 actually do mind that Oct 13 09:31:59 I was thinking "if the phy issue happens, bootrom won't be able to load SPL anyway" but I'm not sure that's true, I don't know if bootrom even communicates with the phy at all.. Oct 13 09:33:34 oh it does, but it actually already auto-detects the phy address **** BEGIN LOGGING AT Sun Oct 13 11:07:24 2019 **** BEGIN LOGGING AT Sun Oct 13 20:05:30 2019 Oct 14 01:06:12 I'm trying to bring up HDMI on a BBB, but I can't seem to get anywhere. My Xorg log is here: http://ix.io/1YzO . What am I missing? Oct 14 02:00:38 HMDI? Oct 14 02:00:51 Sorry. HDMI! Oh. Oct 14 02:01:28 Hey...are you going into /boot and messing w/ the Env.txt file? Oct 14 02:02:20 I saw once on my machine, from my doing, that I disabled, w/ a #, my HDMI usage in that file. Oct 14 02:35:59 Where would I find that file? Oct 14 02:36:08 Oh, /boot. Oct 14 02:36:16 I have no such file there. Oct 14 02:41:21 uEnv.txt is in /boot. Oct 14 02:41:30 Sorry. I forgot the u. Oct 14 02:42:23 set_: Not seeing any such file. Oct 14 02:42:40 If you go to that file and scroll down to where it states... Oct 14 02:42:41 Oh. Oct 14 02:43:06 Did you get an image from the Beagleboard.org/latest-images site? Oct 14 02:43:30 No, I'm using ArchLinuxARM's image. Oct 14 02:43:42 Oh. No wonder I am off base here. Oct 14 02:43:43 I figured someone may have a clue if I gave them the Xorg log in here. Oct 14 02:43:50 I can link you to the image if that'd help? Oct 14 02:44:06 Oh. Wait for another person. I do not use those images at all. Oct 14 02:44:12 I am sorry. Oct 14 02:44:31 I have been lost in /latest-images for years now. Oct 14 02:44:33 set_: No worries - thanks for responding! Oct 14 02:44:49 No issue. Maybe another Arch Linux person is around. Oct 14 02:45:23 arch linux on the beagle? Oct 14 02:46:17 GenTooMan: Yep. Oct 14 02:47:23 Well debian appears to be the most common used one. I haven't used Arch for a while (7 years) although I do like the package system they implemented Oct 14 02:48:13 In your config- file in the /boot directory, do you see anything relating to HDMI? Oct 14 02:48:13 GenTooMan: To each their own - I'm mostly here to figure out the HDMI issues, which I don't think are distro-specific. Oct 14 02:48:28 set_: Are you referring to boot.txt? Oct 14 02:48:59 No. It should be in the /boot dir. and state config- Oct 14 02:49:44 set_: This is 'ls -A /boot': http://ix.io/1YAw Oct 14 02:50:08 koz_: Please use pastebin. Oct 14 02:51:29 set_: https://pastebin.com/mVzakwbn Oct 14 02:53:25 Dang it. I wonder what boot.txt states. I know it is not the .img files we need to look into. Oct 14 02:53:42 zImage is probably elf stuff, right? Oct 14 02:54:13 zImage is most definitely elf stuff. Oct 14 02:54:17 I can paste you the boot.txt. Oct 14 02:54:27 Otay. oh. yea, that would be nice. Oct 14 02:54:50 i will plug in my BBB instead of this BBBW. I forgot this one does not have HDMI. Oct 14 02:54:58 set_: https://pastebin.com/aJP8GHCL Oct 14 02:56:23 Nope. That meant nothing to me. Hmm. Oct 14 02:56:35 Does your init file have access for reading? Oct 14 02:56:50 Sorry, what do you mean by 'init file'? Oct 14 02:57:22 in /boot, there should be some type of init file. Let me check the paste again. Oct 14 02:57:37 initramfs? Oct 14 02:58:13 initramfs-linux.img is a binary. Oct 14 02:58:20 Dang it. Oct 14 02:58:24 I saw the boot.txt file Oct 14 02:58:47 There was nothing in it that would lead me to believe that you set up HDMI when setting up your image. Oct 14 02:59:05 What would you normally need to do for that? Oct 14 02:59:20 Yikes. I would need to look again at my book. Oct 14 02:59:28 Hey. Oct 14 02:59:30 I know. Oct 14 02:59:52 Wait. Let me get this site for you. It will give some basics on setting up images on the BBB. **** ENDING LOGGING AT Mon Oct 14 02:59:57 2019