**** BEGIN LOGGING AT Fri May 15 02:59:57 2020 May 15 04:34:19 wasn't haskell the cool language of the year in 2009? :) May 15 04:37:07 * paulg thought he was the rebel in "Leave it to Beaver".... May 15 05:08:40 paulg: heh had to wikipedia but Eddie is still alive May 15 08:05:42 hello all, i'm tusk during kernel boot at : Failed to execute /init (error -13) any clue ? After checking cpio.gz init is present, link to /sbin/init ... what I have missed ? Thx May 15 08:38:15 Ninic0c0, fs mount problems ? May 15 08:38:32 Ninic0c0, maybe post your boot log somewhere if possible May 15 08:43:13 @kroon not sure https://pastebin.com/t3SSBsit thx for taking a look May 15 08:44:15 Ninic0c0, i think your kernel arguments are b0rked May 15 08:44:30 Ninic0c0, indicated by the line "/bin/sh: earlyprintk: No such file or directory" May 15 08:47:14 hooo i'm stupid or maybe blind... thx for the advice i will double check :) May 15 09:20:21 @kroon not so trivial Kernel command line: console=ttyPS0,115200 mtdparts=nand0:5m(bootloader)ro,256k(env),256k(marker),100m(rootfs1),100m(rootfs2),-(storage) modprobe.blacklist=i2c-mux-pca954x,pca953x seems okay... May 15 09:23:04 is that what the boot log says ? no "earlyprintk" ? no "root" ? May 15 09:24:28 i have remove earlyprintk to test, same issue, no root because i'm using initramdisk May 15 11:14:50 zeddii: awake? May 15 11:27:06 hi guys, I want to use wic to generate an extra empty data partition without increasing my wic image size. Is it possible ? May 15 11:28:14 ebail: as wic images are meant to be bitexact, not really. May 15 11:29:04 ebail: you could create a really small partition (which will only be a small increase) and then resize at first boot. but then, you could also just outright create the empty partition completely upon first startup. May 15 11:29:27 that was my B plan ;) Letothe2nd May 15 11:29:53 better make it plan A :) May 15 11:30:13 Or I could compress my wic image to reduce the size... May 15 11:30:31 i don't really get whats to be gained from it. May 15 12:28:36 Crofton|cloud: hell no. I was asleep May 15 12:28:36 Letothe2nd: using xz I can compress my wic image and reduce the size from 17G to 388M May 15 12:29:23 but it would be useful only for file transfer purpose May 15 12:29:57 I am wondering if genimage (used in Buildroot) could offer this feature May 15 12:32:19 ebail: that sounds like a massive misusage of wic anyways, if your image has got 17G, which seem to consist of only a fraction of real payload. May 15 12:33:08 Letothe2nd: I added 12G of data partition May 15 12:34:48 ebail: well i don't know your usecase and therefore can't really judge. but it sounds very much like you want a massive filesystem hierarchy, yet have little payload to start out. this is not something that binary images handle gracefully by themselves. May 15 12:36:13 not very massive for the moment Letothe2nd. It is X86 based distribution for VM and containers May 15 12:37:15 at least for containers wic is completely unsuited. May 15 12:37:19 machine, distro and images are customized May 15 12:37:49 Letothe2nd: what do you mean by that ? What would you suggest ? May 15 12:38:08 ebail: why not use a plain tarball? May 15 12:38:44 Letothe2nd: because it is not easily flashable ? May 15 12:39:02 what would be the bringing values ? May 15 12:39:14 container engines hardly ever run directly on flash. May 15 12:39:40 I am not runing on a flash. I am using an hard disk May 15 12:40:01 s/flash/rom storage of any kind/ May 15 12:40:06 I am not on a ARM platform :) I am dealing with a big Xeon May 15 12:40:44 granted i certainly do not know them all, but at least each i've seen could either load a filesystem tarball or point in the filesystem hierarchy. May 15 12:41:13 and to transport that, a simple tarball is definitively the superior choice. May 15 12:41:27 VMs are a different beast, thats sure. May 15 12:43:21 Letothe2nd: for a first install, I would prefer a flashable image. Then you can then use swupdate or RAUC to update certain partitions and so you would deal with small archives May 15 12:44:08 for the base os, sure. for a vm, maybe. for a container? no in my world. May 15 12:44:19 but if it suits your needs, why not. May 15 12:45:10 I am not using wic for containers and VM May 15 12:45:23 I am speaking about my host distribution May 15 12:45:52 that was totally not clear from: 12:36 < ebail> not very massive for the moment Letothe2nd. It is X86 based distribution for VM and containers May 15 12:46:18 in that case i would still go for a firstboot script, but hey. May 15 12:47:02 Letothe2nd: sorry if I was not clear. Yep first boot would be the solution May 15 13:36:48 ebail: there's also bmap support available, which might be workable in your usecase, you can write out just the non-zero blocks using bmaptool with wic.xz + bmap file May 15 13:47:11 Hi there, is there an easy way to determine what recipes/layers are setting a particular kernel config? I have a fragment that sets CONFIG_PROC_KCORE=y as well as CONFIG_PROC_VMCORE=y but for some reason in my final kernel KCORE is not set. It seems like something is overriding it. May 15 13:51:17 RP: I forget, do we usually add "master" to the LAYERSERIES_CORENAMES about now? May 15 13:52:26 RP: Asking because I want to make a change to meta-arm that will make it incompatible with dunfell, so putting that in the LAYERSERIES_COMPAT doesn't seem correct May 15 13:54:35 aaronov: bitbake-layers show-appends should point out the files involved in the setting May 15 13:57:38 how can I rename DTBs at the very end of the packaging? I have my own dt-blob.dts which is compiled to dt-blob.dtb, and it's successfully copied to the /boot but it should be dt-blob.bin. the whole build script is only fitted to .dtb and that's what make it pick up the dts and build it May 15 14:25:39 @mckoan: thanks, I was wondering if there was something a bit more specific to config settings. I've grepped a lot of files trying to see where it might be set but it hasn't shown up. Is there anything else that could cause a config value to not be set to what I have in my append layer? May 15 14:29:25 JPEW: we add gatesgarth May 15 14:29:38 JPEW: I keep forgetting to do this, let me sort May 15 14:40:04 JPEW: its in -next, just checking i got it right May 15 14:40:54 @aaronov: can you share your configuration ? I’m currently extending / changing the audit of the final config to allow an explicit run against all configs (not just hardware/BSP ones), and that would show more details, but I’m tied up on finishing some other things right now .. so I can instead help debug. May 15 14:53:06 aaronov: depends on what you are doing to get that May 15 15:05:20 zeddii: dropping those patches did resolve the issue btw May 15 15:05:42 zeddii: you could create a shell function to abstract that code FWIW May 15 15:07:03 I started with a shell function and moved to python after May 15 15:08:21 since those vars have to be in the environment of the function that calls, it felt easier to call the python routine and assign it in the function. versus trying ot call shell function and have the exports work. Or does a called shell function get concatenated into the caller by bitbake ? May 15 15:09:11 I have it the python one looking ok here now anyway. I missed the fact that it would be parsed on a pass and not always when it was called. So things weren’t setup properly on those passes. May 15 15:21:36 zeddii: fair enough. Shell wouldn't be auto concatenated May 15 15:26:08 I setup a no distro build here, and will send a v2 to replace the previous efforts once I’ve also confirmed it on the AB. May 15 15:26:53 once my slow fetch of the kernel works, I’ll know more :D May 15 15:26:58 * zeddii wanders off. May 15 15:43:41 @zeddii - which configuration do you want? May 15 15:46:49 what branch, what kernel, what layers, etc. pretty much what it would take to run the build locally. otherwise, it is just guesses. May 15 15:48:21 I fixed my layer priorities so my debug layer has the highest - that still didn't fix the KCORE config issue. Something keeps setting it to not set. I did more searching and found that a lower layer was setting KERNEL_EXTRA_FEATURES ?= "features/netfilter/netfilter.scc features/security/security.scc". The security.cfg has "CONFIG_PROC_KCORE is not set" in it. May 15 15:48:50 Zeus, Kernel 4.14 (linux-intel) May 15 15:49:18 It's a complicated configuration so it might be hard to give all needed details :( May 15 15:49:52 that’s the thing, it isn’t just that something might be setting it to not set. May 15 15:50:09 it might be a missing depdendency or a wraping Kconfig that is making it unreachable. May 15 15:50:23 Main Public Layers: meta-intel, meta-openembedded, meta-security May 15 15:50:44 if you drop into menuconfig and search on the symbol, you can get a hint as to what options might be missing. May 15 15:51:04 if that fails, I can point you into the configuration qeueue and merge_config logs and you can see if somethign is clobbering it by checking the configfs. May 15 15:51:16 It says it only has a dependency on MMU and PROC_FS, which I forced to =y May 15 15:51:50 and they are showing as =y in the .config ? so in that case, it isn’t the direct depdencies that are the issue then. May 15 15:52:23 The logs might be useful - I was trying to find any log that could help, I had a look at the kernel config check log but nothing mentioned KCORE May 15 15:52:35 Yep MMU and PROC_FS are =y in the config May 15 15:53:31 Can you confirm where the final merged .config lives? I was look under "tmp/work/corei7-64-intel-common-poky-linux/linux-intel/4.14.67+gitAUTOINC+a889c43359_8730ef3292-r0/linux-corei7-64-intel-common-standard-build" May 15 15:54:05 that’s the final .config yes, but it isn’t the merged config per-se, it’s the merged config fed through lkc, so more changes happen after the merged config at that point. May 15 15:54:39 Ok, well the config in that location has KCORE not set May 15 15:55:38 sec. let me look at some zeus things and see if anything jumps out. May 15 15:56:17 Thanks May 15 15:57:18 aaronov: I’m almost sure it is this: recipes-kernel/linux/linux-intel_4.14.bb:KERNEL_EXTRA_FEATURES ?= "features/netfilter/netfilter.scc features/security/security.scc" May 15 15:57:20 how do i create a multilib32 sdk? i thought this is the way to go: bitbake lib32-my-image -c populate_sdk May 15 15:57:30 Also out of curiosity - is there documentation that explains the merge process and where the logs are etc? I had to guess for the location of the final .config - is it documented? May 15 15:57:31 since scurity.scc has: features/security/security.cfg:# CONFIG_PROC_KCORE is not set May 15 15:57:35 but the lib32-my-image target has no populate_sdk task :( May 15 15:57:46 aaronov: we put it all in the kernel dev manual back in the day. May 15 15:57:52 I’m sure it needs revisiting now. May 15 15:58:32 Ok - that is where I ended up and have now set this in my append: KERNEL_EXTRA_FEATURES = "features/netfilter/netfilter.scc features/debug/debug-kernel.scc" May 15 15:58:51 Waiting for things to build now May 15 16:00:13 So somewhere in https://www.yoctoproject.org/docs/3.0/kernel-dev/kernel-dev.html it should specify useful output locations May 15 16:00:18 yup, that should be it .. the PITA of tracking this down is that you do require a kernel config run to be sure, my scripts can do some analysis now, but I’m extendingn them to be easier to use. that’s in master, but the techniques will apply to older branches. May 15 16:00:57 aaronov: yah, I contributed it in the day, but I can’t honestly remember reading the final version. I’ll go check and write some HOWTOs on debugging it as a stop gap measure. May 15 16:01:28 Thanks zeddii May 15 16:02:19 Some tool for tracking where configs get set from would be great. I had to do a lot of digging/grepping to find security.scc reference. May 15 16:03:09 that’s what they do under the covers when running the audit now, but yes, that’s what my updated ones do. easier to invoke in a task (or command line) just to dump details. Saves me documenting it all and making it manual. May 15 16:03:20 Maybe some dependency graph of sorts that can show what touches it and what has the "final say" in it getting to the final config May 15 16:03:49 Awesome, looking forward to the updates May 15 16:04:14 Is that a YP3.2 change? May 15 16:04:18 it’s easier to say than do, but that’s what “symbol_why.py” does in the tools, but it doesn’t take into acount the .cfg’s that fed into it, which is what I’m putting into it now. May 15 16:04:51 yah, it’s what I’m working on for the next release, since I’ve had a couple of pings on it now and debugging was a pain, so I started a bigger set of changes for the audit, which led to the broken out script. May 15 16:05:05 Yeah, I don't assume its an easy task :) May 15 16:05:29 I wasn't aware of symbol_why.py May 15 16:05:41 Is that a user script or something run under the hood? May 15 16:06:25 and the tools themselves already dump the configs, etc, but that’s part of the gathering process, so you can easily get a listing of everything from them, but the arguments aren’t friendly to call by hand. May 15 16:07:04 symbol_why.py is in the kern-tools, and called as part of the audit when a mismatch is found, it will be a user script, it isn’t right now just because I it’ll blow up badly on bad inputs, etc. May 15 16:07:50 my python foo is significantly better now then when I did all that about 3 years ago, so I’m fixing it up finally. May 15 16:08:02 * zeddii notes he’s filling up the channel with rambling. May 15 16:08:42 I like the rambling, better then an inactive channel :D May 15 16:09:08 Also I'm learning some good stuff here that might otherwise be hard to find online May 15 16:10:22 I’ll be lurking on here, but stepping away shortly. if that change doens’t work. ping and I’ll look more, but be assured. I know the debug of this is not nice, and I’m finally working to address that. May 15 16:11:45 Appreciate it May 15 16:12:21 Just looking up the kern-tools repo now as I wasn't really aware of it before May 15 16:14:02 You might be able to recommend what I do with all these messages: https://pastebin.com/e3Rtdxhs May 15 16:14:39 That is just a subset, there is a lot spammed to console and I'm not sure if I should do anything about them May 15 16:16:48 you can ignore them for now, but in there, you can see symbol_why running, etc. May 15 16:17:27 you could create a run-last fragment that set them to the matching values and they’d do away, but if you are ok with the boot you can ignore them. May 15 16:18:10 and that’s the balance I’ve always had to do with the warnings, how many to show, how to prioritize them, some people want them, some don’t .. there’s some variables that will inhibit them from being shown as well, you could crank it down and hide them. May 15 16:18:19 Ok. Yeah no real issue and there is another message "This BSP sets config options that are not offered anywhere within this kernel" May 15 16:18:44 Was just curious if I should do anything about them and what the process would be to properly clean things up May 15 16:18:52 yah. the fragments shouldn’t do that, but they aren’t always fully up to date with the version, or are serving multiple versions, etc. May 15 16:19:45 you’d need to update the kernel-cache version of them, if using the upstream ones (that’s the second repo on the linux-yocto clones). or you’d have to have local versions that took precedence. May 15 16:20:46 RP: just wondering, is there any path to having a scheme for accessing one multiconfig's variables from another multiconfig, or does the current architecture make that a complete no go? May 15 16:22:12 @zeddii: thanks for the info. I'm locked in to kernel version and cache because of the BSP provided by intel for our HW platform - so guess I just ignore the messages May 15 16:23:46 SUCCESS! May 15 16:24:00 cool. May 15 16:24:08 It was security config that was sneaking in May 15 16:24:17 as part of the howo’s i’m working on, I can add a bit about silencing the warnings with local changes. May 15 16:25:18 I still don't understand why KERNEL_EXTRA_FEATURES that is set in a lower priority layer would clobber my explicit =y in a higher layer May 15 16:25:42 smurray: you can't do it today. Its not totally impossible but would have horrible overhead May 15 16:26:03 black magic May 15 16:27:39 RP: that's about what I expected. I was idly thinking that geting the value of DEPLOYDIR and similar would be nice for the image composition usecase May 15 16:28:36 aaronov: that was documented in what I wrote years ago. kernel features are a contract between the meta-data and the kernel config, if they fail to be set they are an error. So they come last in the process. Only things that are really required should be in there. But that’s the part which is not always used right or not obvious). I’ve tweaked it a few times, but I always end up breaking someone’s layer somewher May 15 16:28:44 ok. really stepping away now, need to get food. May 15 16:30:30 Is it possible to do KERNEL_EXTRA_FEATURES_remove? May 15 16:31:03 Thanks - I'll brush up on the doco May 15 16:41:47 yep. it’s just a variable like any other. May 15 16:49:39 Cool. I can probably just remove security rather then clobbering the variable May 15 16:54:22 zeddii, just checked gmail spam folder - had 99, of which 33 were false positives. :-/ May 15 16:55:11 I wonder if this is an intentional ploy to deter people like us with archives of oe/yocto/lkml feeding there.... May 15 17:06:49 smurray: right, you'd really want to configure it to use the same location when writing the multiconfigs (or a known layout) May 15 17:08:18 RP: I've used known layout in my previous tinkerings, as sharing it seems a bit fragile May 15 17:13:11 smurray: I can imagine that being safer, yes May 15 17:38:28 Anyone have experience with configuring kdump? May 15 17:40:48 aaronov, I'd be really surprised if we (yocto) were doing anything specific about kdump, vs. say any other distro. May 15 17:41:27 as such, I'd hope any "generic" online references and how-to guides would be appropriate. May 15 17:42:29 I was hoping for that but I'm having trouble getting it to reboot into crash kernel after I echo c to sysrq-trigger. It just seems to stall the entire system. May 15 17:42:57 I also notice https://linux.die.net/man/5/kdump.conf doesn't seem to match anything in the /etc/sysconfig/kdump.conf that comes with Yocto kdump May 15 17:46:30 I've really never had a use for kexec/kdump personally, but I wouldn't think that sysrq-c should just wedge. May 15 17:47:15 I'd expect it would trigger a NULL deref or a kernel panic in order to test the whole crash path... May 15 17:48:28 are you doing it from a console where you'd have an opportunity to see kernel printk() output - like a uart, with "console=ttyS0,115200 May 15 17:48:34 (or similar...) May 15 17:48:35 Yeah, it causes a NULL deref and I see the OOPS and call stack print out, then it just hangs May 15 17:49:24 RP: I’ve got a fixed python patch. since you tossed the old one, I’m squashing them together and will send a v2 version. May 15 17:49:31 I'm waiting on a serial adapter so I can get a serial console, right now I'm stuck with a monitor connected to the device May 15 17:49:42 can you check the .config and ensure you've got the related options set - KEXEC or whatever they are? May 15 17:50:33 .config is setup correctly and kdump service starts without issue - I was missing PROC_KCORE which @zeddii helped me debug earlier May 15 17:50:58 ok, I missed that dialog. May 15 17:52:13 again, never used it -- not sure if it expects you to use the reboot-on-panic boot arg or any of that stuff. May 15 17:52:34 https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-kernel/kexec/kexec-tools/kdump May 15 17:53:03 I see "Kdump started up" - which suggests all that scripts sanity checks are passing May 15 17:53:33 yah. even being around kdump for > 10 years via CGL, I’ve never used it. May 15 17:55:06 sheesh zeddii - what kind of kernel maintainer are you? May 15 17:55:09 * paulg runs May 15 17:55:09 First time for me :) Running into a complete system stall when resetting a device, which lead me to setting up environment for kernel debugging and get a crash dump to inspect May 15 17:55:59 the configs I had mainly used (with the above comment still implied), were powerpc/arm and hence dtb wrangling was required, at least x86 should be simpler on that front. May 15 17:56:13 assuming the devices are willing to not lock up on the warm restaart May 15 17:56:53 aaronov, what kernel version are you on -- in the off chance that is relevant? May 15 17:57:04 linux-intel 4.14 May 15 17:57:45 ah, so that will be an intel specific patched thing - it is possible they indavertently vandalized sysrq-c May 15 17:58:03 * paulg goes to read code to see what sysrq-c really does. May 15 17:58:50 The only thing different in my setup to every tutorial online, is that I already boot into a "bootloader" kernel and then kexec to a "main" kernel/OS.. so there is more kexec-chaining in my environment May 15 18:00:05 yeah, again you've got me at a knowledge defecit on kexec -- I've NFI just what level of "nesting" it is happy with and/or if there are limitations. May 15 18:00:47 "normally" it is a one shot deal into a crippled "data collection" environment - IIRC.... May 15 18:01:23 at least I think that was the original use case with the CGL stuff from 2005-ish that zeddii referenced above. May 15 18:05:53 Yeah that is the general idea May 15 18:06:31 Think I'm stuck till I get the serial console - I have a feeling its the cmdline being passed into kexec that needs adjusting. It might not be finding the kernel May 15 18:11:36 all I can suggest, is go test what you think should work on a COTS x86-64 system, with a generic (yocto/mainline) kernel and verify all your steps and crashkernel= bootarg usage are sane and function as documented. May 15 18:12:33 And maybe that will shed some light on what you can do/try/check on the more complex system doing the "double bounce" kexec thing. May 15 18:13:13 It is always easier to learn how to ride a moped than learning how to ride a GSXR1000 :) May 15 18:14:58 Haha good point. What do you ride? May 15 18:16:12 mostly dirt bikes. May 15 18:17:46 I just started riding bit over a year ago, no dirt bikes yet May 15 18:20:09 paulg: I didn't know that! I do a bit of that too... May 15 18:20:12 hey uh May 15 18:20:36 if I add linux-firmware-foo into MACHINE_ESSENTIAL_EXTRA_RDEPENDS , the entire linux-firmware gets installed into the image May 15 18:20:41 that's not right, is it ? May 15 18:20:59 I only need one specific package from the linux-firmware (linux-firmware-foo), not the whole thing May 15 18:34:18 RP, you actually still have free time for hobbies that yocto hasn't yet eaten up? :-P May 15 18:55:43 paulg: not often but sometimes. I prefer the mtb these days tbh but the KTM is fun May 15 18:59:14 I've rode two KTM but never owned one. 990 Adventure, and 510 (520? 525? -- keeps getting renamed) EXC May 15 19:00:35 I'd probably die if I had the 990. Feels like a 100HP dirt bike, until you come into a corner waaay faster, and 200lbs heavier than a "normal" dirt bike. May 15 19:06:07 ah, because I;m installing linux-firmware-dev, which pulls in the linux-firmware , now it makes sense May 15 19:06:59 but I don't care who you are - steering with the rear wheel in any gear at any speed on the 990 will put a shit-eating grin on even the most curmudgeonly soul. May 15 19:08:26 (dirt roads in Fumbuck Nowhere, of course....) May 15 19:13:15 paulg: I had a 990 appear at the top of an incline whilst I was trying to put some marker posts/tape back up to slow the riders down. Small problem was it was on its side and heading for me. Not something I care to think about given the weight! May 15 19:13:34 paulg: mine is a 300EXC, quite a different machine, I like my 2Ts May 15 19:19:58 heh, last two-smoke I owned was a 1982 IT 175-J --- sold it in 1987 maybe? Got sick of mixing gas. May 15 19:25:35 and man, did it eat gas. May 15 19:26:30 er, petrol, for RP 's benefit. ;-) May 15 19:35:18 * zeddii waits for his build to get scheduled. May 15 19:40:33 zeddii, I'll get workers back up for it ASAP. May 15 19:42:23 paulg: I can translate although an LPG conversion on a two stroke would be interesting :) May 15 19:42:33 no worries. I was hoping it wasn’t my fault :D May 15 19:48:27 zeddii, It's running https://autobuilder.yoctoproject.org/typhoon/#/builders/47/builds/1923 May 15 19:48:44 woot. I see the tab spewing to life in the background. thanks! May 15 19:54:15 RP, I'm sure rippin thru the forest with a BBQ tank strapped to your back wouldn't upset your rhythm at all. May 15 19:59:32 RP, found a UK version of the IT-175 - I love y'all for embracing the farce of selling them as road legal by sticking on turn signals and a lic. plate mount. May 15 19:59:39 https://www.flickr.com/photos/tblazier/20707335515 May 15 20:00:42 Aussies did (do?) the same - back when I lived there in the 1990s, Honda XRs were sold as "street legal". May 15 20:01:47 Over here in Kanuckistan you had to jump through hoops to get similar XR on the road, via "dual-sport conversion". Royal PITA. May 15 21:41:37 paulg: I have a YZ250 which fits that description quite well :) May 15 21:42:00 paulg: you don't need turn signals May 15 21:42:28 YZ250F or legacy 2T ? May 15 21:43:23 paulg: legacy, its old May 15 21:44:04 is it liquid cooled at least? ;-) May 15 21:45:25 paulg: yes! :) May 15 21:46:05 well then - you might get a free pass on "legacy" May 15 21:47:31 * paulg still rides air cooled dinosaurs. May 15 21:49:01 change the oil, add fuel, and shit keeps working. May 15 21:50:01 * paulg is a dinosaur May 15 21:51:52 paulg: The YZ is 2002 and I have older. I just like crazy two strokes May 15 21:53:31 I find it kinda funny that the hi-end 4T now have higher maintenance requirements than the "old" two-smokes. May 15 21:54:28 "Change the piston after X hours or your crankcase will be showered with aluminum skirt fragments" May 15 21:54:44 old two smokes never grenaded May 15 21:55:08 They just got louder and louder as the piston rattle increased. May 15 21:56:43 Here I am riding a ninja300 :D May 15 21:57:22 Ready to upgrade to a 650 May 15 21:57:51 paulg: I managed to destroy a lot of air cooled YZ125s and CR250s on karts May 15 22:03:43 well, at least there weren't rocker arms, or cam chains or bucket shims to get swept up in the destruction. May 15 22:08:58 * paulg giggles at how a gixxer reference derailed #yocto May 15 22:20:22 paulg: I've been trying not to say too much on the subject too :) May 15 23:28:08 hello guys!! I'm new to yocto and linux development, so please excuse my newbie question. May 15 23:28:56 I'm trying to compile a kernel module that I wrote my own using the toolchain that yocto exposes through the SDK May 15 23:29:27 however, I get an error like this: "fatal error: linux/init.h: No such file or directory " May 15 23:30:00 is there something special in building a kernel module with the toolchain? Am I missing something here? May 15 23:41:04 well, I went further and found an example in poky. However, I tried to compile an example base on a same Makefile as the one in the hello-mod recipe, but got another error, like May 15 23:41:10 make -C M=/home/nucatus/octokmod/module May 15 23:41:10 make[1]: *** M=/home/nucatus/octokmod/module: No such file or directory. Stop. May 15 23:41:10 Makefile:6: recipe for target 'all' failed May 15 23:41:12 make: *** [all] Error 2 May 15 23:41:45 that directory is saying it doesn't exist is actually the pwd of the module May 16 00:32:44 q May 16 00:32:47 exit May 16 00:32:51 \q May 16 01:21:08 apt get install emacs May 16 01:21:47 just sayin.... **** ENDING LOGGING AT Sat May 16 02:59:57 2020