**** BEGIN LOGGING AT Mon Jan 01 03:00:00 2018 Jan 01 07:35:40 Morning! Jan 01 07:35:45 Happy New Year! Jan 01 07:36:17 elvispre: I keep getting the 13 section mismatches, even though I added CONFIG_NO_ERROR_ON_MISMATCH=y to my defconfig Jan 01 07:36:24 Did you do something else locally? Jan 01 08:38:15 Happy New Year Jan 01 08:38:45 Herrie: I added that directly to a Makefile. Jan 01 08:40:51 Herrie: Like so https://github.com/corpuscle/android_kernel_oneplus_onyx/commit/863303bcab6eec67d29e28f129d33ecfd57ea3bd Jan 01 08:41:45 Herrie: Very hacky! Jan 01 08:51:05 Ah ok Jan 01 08:56:57 There doesn't seem to be anything proper from what I can tell :S Jan 01 10:17:06 Morning folks! And happy new year :) Jan 01 12:22:45 elvispre: Somehow your branch works, mine doesn't :P Jan 01 12:34:14 elvispre: Ah OK found it Jan 01 13:18:51 elvispre: Seems this would be a proper fix: https://github.com/XPerience-AOSP-Lollipop/android_kernel_motorola_ghost-OLD/commit/10318586d950520841e1eb3f6f817643d5d2dabc Jan 01 13:52:07 Herrie: "static extern"? (Last file) Not convinced that's much better than the patches I found. (Thanks though.) Jan 01 13:58:40 Currently running a local LineageOS build for my onyx. If that works (and installs and works) then I have a basis for a bit of compare and contrast. Jan 01 14:05:32 elvispre: Well the one you have comments the whole code out :P Jan 01 14:11:11 Herrie: True. Phones don't tend to have a mouse, so I figured it wasn't a big risk :-) Jan 01 15:08:52 elvispre|s: Also true Jan 01 15:08:54 But it's hacky Jan 01 15:08:57 Let me try this one Jan 01 15:28:44 That works Jan 01 15:29:13 So we just have the hacky "CONFIG_NO_ERROR_ON_MISMATCH=y". Not sure this does any harm.... Jan 01 15:29:18 Tofe: You know anything about that? Jan 01 15:30:03 Tofe: This is what we get in the logs: https://bpaste.net/show/b592afd3fa98 Jan 01 16:48:56 elvispre: This might seems like proper fix to not use the CONFIG_NO_ERROR_ON_MISMATCH: https://gist.github.com/luca020400/acaf3d3350eab151ecf2 Jan 01 16:53:18 elvispre: Though it does not seem to address all, but quite some already Jan 01 19:28:55 Herrie: mmh let me see Jan 01 19:29:37 ok, so yes, don't use CONFIG_NO_ERROR_ON_MISMATCH, or you might get crashes on boot Jan 01 19:30:35 Tofe: Guess what I get! Jan 01 19:30:44 :) Jan 01 19:32:15 It looks like someone forgot to put the correct section prefix for some data or function (like __init, __initdata, etc) Jan 01 19:37:22 My local LineageOS build installs and runs OK => I am not totally incompetent and we have some kernel code that definitely does work on the device. That gist Herrie found looks interesting. I'll try that next. Jan 01 19:41:39 well, for instance, https://github.com/corpuscle/android_kernel_oneplus_onyx/blob/cm-14.1/arch/arm/mach-msm/krait-regulator.c#L1412 here you have a reference to a __devinit function ( https://github.com/corpuscle/android_kernel_oneplus_onyx/blob/cm-14.1/arch/arm/mach-msm/krait-regulator.c#L1203 ) without setting any proper variable prefix Jan 01 19:43:20 I'm a bit surprised that the Android build doesn't complain about that Jan 01 19:46:06 Tofe: Oh, that's not what's working. I mean I just built the current LineageOS onyx target using *their* sources. Jan 01 19:46:08 But given the warnings this is happening quite often, so it could be a bit tedious to fix it Jan 01 19:46:39 ah ok, were would the corresponding source be? Jan 01 19:48:21 Got it -- but it's the same code for that file Jan 01 19:48:36 Tofe: I forked Halium which forked LineageOS, all within GitHub. Jan 01 19:48:58 Tofe: OK, so your comment stands. Jan 01 19:51:08 it could be because of this: https://github.com/LineageOS/android_kernel_oneplus_onyx/blob/cm-14.1/include/linux/init.h#L332 Jan 01 19:52:41 do you have MODULE or HOTPLUG enabled? And were they enabled on LineageOS' build? Jan 01 19:53:09 Tofe: Nice find. Jan 01 19:53:10 (@see resp. defconfig) Jan 01 19:55:56 Tofe: You over-estimate where I have got to. I just dropped my kernel source URL into the bb recipe for our build chain. I did no configuration that is not built in. Similarly, I did not tweak the LineageOS build at all. Jan 01 19:56:30 ah ok - well that surely means you use the same defconfig ;) Jan 01 19:57:06 and I think we use a newer binutils than the android toolchain Jan 01 19:57:35 I see few defconfig files in the LOS kernel source. None of the arch folders look like they match this CPU. Jan 01 19:58:00 Perhaps the defconfig is just 'def'. Jan 01 19:59:26 should be https://github.com/LineageOS/android_kernel_oneplus_onyx/blob/cm-14.1/arch/arm/configs/lineageos_onyx_defconfig Jan 01 20:01:10 you could try setting https://github.com/corpuscle/android_kernel_oneplus_onyx/blob/cm-14.1/arch/arm/configs/lineageos_onyx_defconfig#L131 to =n Jan 01 20:01:28 Tofe: Ah yes, thanks. "find . -name '*defconfig"! Jan 01 20:03:26 Tofe: Turn off HOTPLUG to see if that let's the build succeed? Jan 01 20:03:38 yup Jan 01 20:03:54 Good stuff. Jan 01 20:04:09 Back in a bit. Jan 01 20:04:18 though it's a long shot here Jan 01 20:53:31 elvispre: Android/LineageOS/CM use old toolchain to build (GCC 4.7/4.8) you won't have issues there Jan 01 20:53:36 We use new GCC 6.3 ;) Jan 01 20:53:41 That's why we have issues Jan 01 20:54:12 elvispre: You might want to run the mer-kernel options script from Halium Jan 01 20:54:23 That should give you what needs updating in defconfig Jan 01 21:21:20 elvispre: As per http://docs.halium.org/en/latest/porting/build-sources.html#modify-the-kernel-configuration Jan 01 21:21:24 That would be needed anyway Jan 01 21:29:56 Herrie: Thanks. Jan 01 23:50:36 Herrie: All those gcc-4.9.patch patches are already in place. They are probably a clue though. Jan 02 00:04:59 Tofe: Turning off HOTPLUG does get us past the section mismatch build blocker. **** ENDING LOGGING AT Tue Jan 02 03:00:01 2018