**** BEGIN LOGGING AT Mon Aug 01 02:59:57 2022 Aug 01 08:21:39 Morning! Aug 01 08:24:26 Morning! Aug 01 08:25:04 Herrie: that regulator, we have it, haven't we ? Aug 01 08:26:49 Tofe: Yeah seems so Aug 01 08:44:48 I found a few other defconfig flags, but my TP's were all empty somehow Aug 01 08:44:52 So put them to charge overnight Aug 01 08:46:34 It are these flags: https://bpa.st/HKPA Aug 01 08:48:58 we should have ULPI already, I hope$ Aug 01 08:49:13 MSM_OTG too Aug 01 08:49:37 for the rest I don't know if they influence our issue Aug 01 08:51:14 insteresting: # CONFIG_USB_ULPI is not set Aug 01 08:51:54 I found these in the default qcom one Aug 01 08:52:10 Did some side by side comparison of the various defconfigs Aug 01 08:52:25 Not sure they will be of any help, but worth a try Aug 01 08:53:00 we should also commit the savedefconfig for our current defconfig, too. It'll help figure out what is missing... Aug 01 08:53:16 let me see if I manage to do that Aug 01 09:04:23 mmh can't reach my builder, this will have to wait Aug 01 09:17:22 Morning Aug 01 09:18:19 I can do that :) Aug 01 09:20:20 be aware that in some older kernels the kconfig stuff was a bit broken and using defconfig from savedefconfig produced slightly different .config file than the original, I haven't seen it with newer kernels but sometimes it's interesting to compare original .config and then the one used after switch Aug 01 09:23:24 JaMa: Yeah I remember we had some bits in certain recipes to force values Aug 01 09:44:08 done in jansa/kirkstone branch, but there were many unexpected changes in resulting .config file Aug 01 09:44:42 well maybe that's why we have an unexpected result on the device :p Aug 01 09:46:11 how were you updating the defconfig? with menuconfig or by hand? Aug 01 09:46:35 by hand Aug 01 09:46:48 ah, yes, that might be the reason :) Aug 01 09:46:52 I began with taking mbd98's defconfig as-is Aug 01 09:47:06 try bitbake -c menuconfig virtual/kernel Aug 01 09:48:06 the issue is that if some option you've changed to Y depends on something which isn't set or is set to M, it will be changed in resulting .config (and too often the transitive dependencies of kernel options aren't very clear) Aug 01 09:49:13 while menuconfig either won't allow you to change it to Y if the option isn't shown as well or it will update all dependencies to match the Y when you change it on leaf Aug 01 09:56:11 JaMa: I noticed that the Dragonboard developer used some kernel script to do that Aug 01 09:56:19 Is there a way to do the same in Yocto automated? Aug 01 09:56:29 do what exactly? Aug 01 09:57:10 Seems to run scripts/config and enable some features: https://dflund.se/~triad/krad/makefiles/apq8060.mak Aug 01 09:57:16 $(CURDIR)/scripts/config Aug 01 09:57:28 --enable ARCH_MSM8X60 etc Aug 01 09:57:45 Which seems to enable the defconfig options and dependencies as well, similar to manual method with menuconfig Aug 01 09:58:14 I don't think yocto has any specific support for this, but maybe you can just call the same inside devshell Aug 01 10:00:32 JaMa: I also couldn't find it, just it seems easier vs menuconfig manually Aug 01 10:02:48 there are various ways to edit the kernel config (menuconfig, xconfig, etc), I'm not very surprised to see a shell approach is available Aug 01 10:03:00 it looks quite convenient too, I must say Aug 01 10:04:14 if it works in devshell, then it would probably work in shell task as well (e.g. do_configure) Aug 01 10:04:52 and I even think this is how the handling of partial defconfig should have been implemented in yocto, instead of bluntly patching defconfig Aug 01 10:05:05 but to me it looks wrong to store static defconfig and then modify it every time in do_configure, IMHO better to modify it with any tools you like (including this scripts/config) and then store the final copy (preferably minimalized with savedefconfig) Aug 01 10:06:23 JaMa: well, there are some usecases, like activating the right options for halium or for usb gadget debugging Aug 01 10:06:50 or maybe this could be a way to check the defconfig, instead of modifying it Aug 01 10:07:12 you mean to enable/disable these based on some bitbake variable? Aug 01 10:07:39 or based on a inherit Aug 01 10:08:01 I think we had (probably still have) in kernel.bbclass, but it never got used very much Aug 01 10:08:46 yes, we still have it; we also have kernel_android.bbclass which deploys a fastboot image Aug 01 10:08:54 and it was long time before linux-yocto, so probably some sed like in kernel_conf_variable_fixup and similar grep for check :) Aug 01 10:11:39 I was thinking about kernel_conf_variable in meta-oe/recipes-kernel/linux/linux.inc Aug 01 10:12:20 but that's indeed very limited, but simple Aug 01 10:30:39 JaMa: I thought savedefconfig would strip out all those usused values with # ? Aug 01 10:31:48 Herrie: I'm glad if it doesn't Aug 01 10:37:33 Tofe: Yeah it's helpful to have them on 1 hand as well of course Aug 01 10:37:41 Herrie: not only unused, but also the ones matching the default (and will keep those explicitly not set with # where the default is Y or M) Aug 01 10:38:14 JaMa: Ah clear Aug 01 11:45:04 it's alive! moboot 0.3.8 on my TP after 4 days of charging :) Aug 01 11:47:06 JaMa: OK great! Aug 01 11:47:19 Well I released 0.4.0b2 which should work with mainline kernel ;) Aug 01 11:47:51 https://github.com/Herrie82/moboot/releases/tag/moboot-0.4.0b2 Aug 01 11:48:13 Tofe added some fixes there, I added shortloin (Go) support Aug 01 12:46:41 should we add recipe for novacom-native from https://github.com/openwebos/novacom ? Aug 01 12:47:26 my old gentoo ebuild used palm-novacom_1.0.80_amd64.deb which isn't available anymore on https://cdn.downloads.palm.com/sdkdownloads/3.0.4.669/sdkBinaries/ and I would rather build it locally Aug 01 12:49:19 OK no objections Aug 01 13:12:32 JaMa: I guess we'd need to pack it properly for systemd Aug 01 13:12:42 At least I needed to do some tweaks for KUbuntu for it to work Aug 01 13:12:51 Since the one I could find was still using upstart Aug 01 13:14:42 dkirker posted the .service, so we could fork and add it or simply add a patch to the recipe: https://forums.webosnation.com/hp-touchpad/331153-novacom-drivers-ubuntu-64bits.html#post3449178 Aug 01 13:15:51 that's for novacomd, do we need novacomd on host? Aug 01 13:16:40 ah I guess we need both Aug 01 13:16:40 $ tmp-glibc/deploy/tools/novacom Aug 01 13:16:41 failed to connect to server Aug 01 13:21:32 Well I needed to make a backup of a Go with webOS 3.0.6 (used to be from a former HP/LG employee) who updated it using internal server ;) Aug 01 13:21:47 So I pulled the image, needed novacomd for that in order to use meta-doctor Aug 01 13:26:24 with novacomd running (from 1.0.80) it's not much better: [2022/8/1 14:25:30] novacom_usb_rx_thread:413 -- usbll(00000000) error: reading packet, result(-1), errno 9 Aug 01 13:26:27 [2022/8/1 14:25:30] novacom_usb_rx_thread:434 -- usbll(00000000) reading packet, reads(2), duration(100ms), result(-1), last_errno 9 Aug 01 13:26:30 will check a bit later Aug 01 13:27:40 [ 7887.328885] usb 9-4.3: new high-speed USB device number 6 using xhci_hcd Aug 01 13:27:40 [ 7887.435044] usb 9-4.3: New USB device found, idVendor=0830, idProduct=8071, bcdDevice=10.00 Aug 01 13:27:43 [ 7887.435052] usb 9-4.3: New USB device strings: Mfr=1, Product=2, SerialNumber=0 Aug 01 13:27:46 [ 7887.435055] usb 9-4.3: Product: Palm Aug 01 13:27:48 [ 7887.435058] usb 9-4.3: Manufacturer: bootie Aug 01 13:30:46 That looks OK Aug 01 13:58:39 even the errors reading packets? Aug 01 14:58:27 JaMa: I guess those errors not Aug 01 14:58:32 Could be the cable? Aug 01 15:25:33 it happens with 3 different cables, will try it will laptop later Aug 01 15:27:37 JaMa: OK well the TP is quite picky with cables Aug 01 15:28:02 I have many that simply don't work with it at all Aug 01 17:07:35 works with the same cable from laptop Aug 01 17:07:38 ] battery status Aug 01 17:07:38 battery present Aug 01 17:07:38 voltage: 3538000 Aug 01 17:07:38 percentage: 29 Aug 01 17:07:38 current: -537968 Aug 01 17:07:40 capacity: 1507200 Aug 01 17:09:42 hmm and went bad again after chg detect Aug 01 17:09:43 [2022/8/1 18:04:07] dev '7c7eb1b9e425de602a071220bdaa6f456f8bd3f0' via usb type topaz-bootie Aug 01 17:09:46 [2022/8/1 18:04:40] novacom_usb_rx_thread:413 -- usbll(00010008) error: reading packet, result(-1), errno 108 Aug 01 17:09:49 [2022/8/1 18:04:40] novacom_usb_rx_thread:434 -- usbll(00010008) reading packet, reads(2), duration(100ms), result(-1), last_errno 19 Aug 01 18:14:36 JaMa: Hmmz weird Aug 01 18:14:57 Tofe: On the USB not working, could this be because I enabled serial console by default on Moboot somehow? Aug 01 18:15:10 Just a thought that came to mind just now Aug 01 18:33:26 Herrie: I don't know, but I hope not Aug 01 18:35:42 mmh where do I find savedefconfig again ? Aug 01 18:35:50 Tofe: I also hope not was just a thought Aug 01 18:35:50 I see a .config, but that's all Aug 01 18:36:13 Because it does enable GPIO58 I think if I recall correctly Aug 01 18:36:57 well I'm really not great with these GPIO thingies Aug 01 18:43:24 I just saw CONFIG_REGULATOR_QCOM_USB_VBUS, should it matter ? Aug 01 18:44:18 Tofe: -c savedefconfig shows its location, see my commit where I have included the output Aug 01 18:44:46 ah, yes, thanks :) Aug 01 18:44:47 https://github.com/shr-distribution/meta-smartphone/commit/b1c3bf5541a0889916c995aa7212d9343d8b7ba5 Aug 01 18:58:49 Tofe: Well based on this patchset it seems to be USB C related, so not something we should worry about? https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg2194085.html Aug 01 18:59:33 But maybe something similar exists for the micro USB, not sure Aug 01 19:10:57 I'm just going through the diff between our defconfig and the resulting .config Aug 01 19:12:12 JaMa: looks like savedefconfig is broken nowadays: scripts/Kconfig.include:56: Sorry, this linker is not supported. Aug 01 19:12:28 I guess the gold linker patch doesn't cover this usecase Aug 01 19:16:29 you need to cherry-pick the fix to oe-core kirkstone Aug 01 19:16:46 https://git.openembedded.org/openembedded-core/commit/?id=f0c9e1bf35289afa63f040bbc13ca1755e4db0f9 Aug 01 19:17:03 oh, thanks! Aug 01 19:19:18 yes it works much better now :) Aug 01 19:21:32 "CONFIG_USB_CHIPIDEA_MSM is not set" ? looks surprising Aug 01 19:29:57 In the end, I'm not that fond of savedefconfig: we don't have the context of the context for the various options, and we don't have the list of all available options Aug 01 22:39:25 Tofe: For better debugging we might need: https://github.com/torvalds/linux/blob/7098cff2a3f4db5f844607b8a09bd191618c769e/arch/arm/Kconfig.debug#L371 ? Aug 01 22:40:19 And these 2 values: https://github.com/torvalds/linux/blob/7098cff2a3f4db5f844607b8a09bd191618c769e/arch/arm/Kconfig.debug#L378-L379 Aug 01 22:40:37 That's via serial, but might give some more low level info there via serial **** ENDING LOGGING AT Tue Aug 02 02:59:57 2022