**** BEGIN LOGGING AT Mon Dec 12 03:00:00 2016 Dec 12 03:33:38 I'm trying to get a ralink wifi dongle working. I added linux-firmware-ralink, what else do I need to do. lsusb shows the device, lsmod shows rt2800usb and rt2800lib, but iwconfig shows no interfaces. Dec 12 03:34:41 also /lib/firmware contains rt2860.bin, rt2870.bin Dec 12 03:35:08 steven2: anything interesting in dmesg? Dec 12 03:36:25 When I hot-plug the device I see this "PM: am33xx_prepare_push_sram_idle: EMIF function copy failed" which is the only thing I see that looks odd Dec 12 03:36:32 I'm not sure what that means Dec 12 03:36:54 I'm using the ti kernel tree on a am335x board Dec 12 03:37:01 nor am I. but it doesn't sound especially good Dec 12 03:52:46 Found this in dmesg also, not sure if it is releated: "Direct firmware load for am335x-pm-firmware.elf failed with error -2" Dec 12 04:03:01 I'm trying to understand how/why I might use toaster... is it just an attempt at building a gui over oe Dec 12 04:04:32 It doesn't seem to add anything. Dec 12 04:08:05 @deepwater, it struck me as a tool for exploring recipes for beginners and not a very practical tool for most workflows Dec 12 04:09:22 @bluelighting, looks like my device is not recognized because it's not an alias for it in modinfo Dec 12 04:09:37 several devices for the same vendor, just not this particular part Dec 12 04:13:45 @steven2; I wasn't sure if toaster was widely used by others. I'm struggling with how I can make use of it Dec 12 04:15:09 @steven2, I was kinda hoping I could use toaster for nightly builds, and releases. Dec 12 04:16:05 I played with it for about half an hour and decided it wasn't what I was looking for, but ymmv Dec 12 04:17:16 Well thanks for another opinion, I figured I must have been doing something wrong. Dec 12 07:16:13 hello Dec 12 07:16:34 i'm trying to build a project that needs ~ 50 GB of disk space Dec 12 07:16:48 on centos 7 with the default xfs Dec 12 07:17:41 and i get: WatchManagerError: add_watch: cannot watch /home/cornel/Downloads/test2/meta-openembedded/meta-networking/recipes-support/tcpdump WD=-1, Errno=No space left on device (ENOSPC) Dec 12 07:17:56 but i have 164 GB frree according to df -h Dec 12 07:18:31 /dev/mapper/centos-home on /home type xfs (rw,relatime,seclabel,attr2,inode64,noquota) Dec 12 07:19:09 is there a known issue with xfs on centos 7? Dec 12 07:22:31 c0rnel: could perhaps be that you ran out of inodes? Dec 12 07:23:11 bluelightning, could be this: http://xfs.org/index.php/XFS_FAQ#Q:_Why_do_I_receive_No_space_left_on_device_after_xfs_growfs.3F Dec 12 07:23:52 could be yes Dec 12 07:25:28 bluelightning, if this is confirmed, then maybe the Yocto documentation should say it, if it says Centos is supported Dec 12 07:26:07 but let me re-test this, after mount -o remount Dec 12 07:27:18 c0rnel: this would be specific to the XFS filesystem though Dec 12 07:27:38 bluelightning, exactly Dec 12 07:28:04 right but what I'm getting at is not everyone who is using CentOS will be using XFS Dec 12 07:28:05 i mean, if there's a known limitation, then it's better to know it before you install centos with default options :) Dec 12 07:28:11 bluelightning: yeah. we would need similar help for people running into metadata issues with btrfs, those who run into out-of-inodes on ext4... Dec 12 07:28:30 those who run into casesensitivity on hfs Dec 12 07:28:56 LetoThe2nd, all this problems ar known but not documented? Dec 12 07:29:48 c0rnel: what i'm saying is that filesystem specific problems are well, filesystem specific. i think its not the responsibility of any application that basically just wants to store data to document all the quirks there. Dec 12 07:30:04 LetoThe2nd, i understand this Dec 12 07:30:15 c0rnel: do you think microsoft word comes with documentation on ntfs limitations? Dec 12 07:30:30 still i think a known limitations page referenced would help a lot of guys trying to create a yocto build Dec 12 07:30:57 LetoThe2nd, i would not definitely compare yocto comunity with microsoft 'support' Dec 12 07:30:59 :) Dec 12 07:31:23 c0rnel: once the application really makes use of fs specifics, like docker using btrfs subvolumes or snapshots, then documentation is due. once the application only relies on the fs to provide reliable storage, its the fs documentation duty. Dec 12 07:31:43 sure Dec 12 07:31:53 i don't disagree. LetoThe2nd Dec 12 07:32:00 c0rnel: so i'd rather say the centos guys should document that drawback on their default options ;-) Dec 12 07:32:14 LetoThe2nd, i agree with this also Dec 12 07:32:26 so much agreement on a monday morning, wow Dec 12 07:32:30 :) Dec 12 07:34:08 LetoThe2nd, i will retest and create a ticket on centos side if the xfs problem is confirmed Dec 12 07:34:43 unfortunately this means i have to reboot, bitbake seems usntoppable even after many ^C lines Dec 12 07:36:07 thank you for supprot Dec 12 07:36:10 bye Dec 12 07:36:50 bluelightning: OTOH, i think it might make sense to have some kind of "storage selection" guide on the OE wiki. the recurring question of ram/ssd/hdd could go there too, and in that context a like to this might be sensible. what do you think? Dec 12 07:36:59 s/like/link/ Dec 12 07:37:39 LetoThe2nd: that sounds like a great idea Dec 12 07:38:08 there are tuning parameters that should help performance that could be documented there as well Dec 12 07:38:32 bluelightning: let me see if i can find enough caffeine.... erm, time to do that. Dec 12 07:39:53 * LetoThe2nd thinks he successfully came across as the grumpy old "other peoples problems" guy :-) Dec 12 08:42:56 from time to time i have this problem: a package that was already builded is trying to build again and it fails Dec 12 08:43:38 can i configure yocto to check if the package was already built and if yes, to do a cleanall package first? Dec 12 08:49:49 hello again Dec 12 08:50:11 i try same buid on ext4 partition and again i get no space left on device Dec 12 08:50:22 i also have space there Dec 12 08:50:38 could it be a /tmp problem wrongly reported? Dec 12 08:51:31 or maybe it's stil using files from $HOME (which is xfs)? Dec 12 08:52:36 i also have inodes Dec 12 08:54:53 good morning Dec 12 09:12:07 c0rnel: this is a real hard drive, and not inside a vm? like the container hd is full, but vm hd not? Dec 12 09:12:43 LetoThe2nd, real seagate constellation hard disk Dec 12 09:13:49 c0rnel: well if there is space and inodes free then i'd guess that some path still points to a place that you do not expect Dec 12 09:14:30 LetoThe2nd, ok, how can i find out what is that path? the error only points to the build parent directory Dec 12 09:14:36 as reported before Dec 12 09:14:59 c0rnel: what does err out? a download? a build? Dec 12 09:15:40 LetoThe2nd, let me look deeper into the log file Dec 12 09:17:36 LetoThe2nd, pfff, i don't see in the log who's complaining Dec 12 09:17:45 maybe it's the toplevel process, bitbake Dec 12 09:17:51 c0rnel: that makes it hard to give advice.... Dec 12 09:18:16 sure, i've seen also something else: ERROR: No space left on device or exceeds fs.inotify.max_user_watches? Dec 12 09:18:21 let me check this first Dec 12 09:19:07 c0rnel: if in doubt, pastebin the failing build log Dec 12 09:22:06 thank you LetoThe2nd let me check what i get from this additional info Dec 12 09:24:19 LetoThe2nd, fs.inotify.max_user_watches is 8192. should this value be increased? Dec 12 09:25:46 * c0rnel is searching the internet for this info Dec 12 09:30:34 c0rnel: no idea, i've never had that error Dec 12 09:35:08 LetoThe2nd, it seems this is not the problem Dec 12 09:35:32 exactly same value it has on a machine that builds fine Dec 12 09:39:28 LetoThe2nd, http://paste.fedoraproject.org/505232/35514148/ Dec 12 09:45:19 c0rnel: no idea, sorry. this really a totally vanilla box, no docker,lxc,vbox, wahtever involved? any limits on ram/cpu? all filesystems have free space? Dec 12 09:46:58 LetoThe2nd, fresh ventos 7 install, 16 GB RAM, core i7 processor Dec 12 09:47:24 LetoThe2nd, real machine, not a virtual one Dec 12 09:48:05 /tmp is a little slow, imo: 1.6 GB total Dec 12 09:48:14 s/slow/low Dec 12 09:48:34 c0rnel: hm Dec 12 09:49:32 LetoThe2nd, actually not /tmp, but /run/user/1000 Dec 12 09:49:51 sorry, i misread the log Dec 12 09:50:04 ^about /tmp^ Dec 12 09:51:54 c0rnel: Whats /run/media/cornel mounted as? Dec 12 09:52:32 $ mount | grep media Dec 12 09:52:32 /dev/mapper/fedora-home on /run/media/cornel/22503de9-da5f-4596-ae81-5bee85b92f1d type ext4 (rw,nosuid,nodev,relatime,seclabel,data=ordered,uhelper=udisks2) Dec 12 09:52:40 nrossi, ^ Dec 12 09:52:58 but why /tmp is not tmpfs, i wonder ? Dec 12 09:53:02 c0rnel: Is "22503de9-da5f-4596-ae81-5bee85b92f1d" your build disk? Dec 12 09:53:14 nrossi, yes Dec 12 09:53:29 "/run/media" is not tmp just fyi Dec 12 09:53:48 nrossi, yes, i know Dec 12 09:56:57 c0rnel: have you got two disks/partitions, / and /home? and is /tmp on / or remapped? Dec 12 09:57:25 nrossi, yes, / and /home Dec 12 09:57:33 nrossi, /tmp is on / Dec 12 09:58:13 c0rnel: and so "df -h /" shows a decent amount of free space? Dec 12 09:58:39 nrossi, 47G free space Dec 12 09:59:04 161G on /home Dec 12 09:59:33 136G on media Dec 12 10:00:58 c0rnel: Given the inotify message, what is your system setup for "cat /proc/sys/fs/inotify/max_user_watches" Dec 12 10:01:07 8192 Dec 12 10:01:16 but let me recheck Dec 12 10:01:35 c0rnel, cornel: that sounds low? Dec 12 10:01:57 yep, 8192 Dec 12 10:02:21 nrossi, it's the same on other machines where the build works fine Dec 12 10:02:33 and other operating systems Dec 12 10:03:30 c0rnel: Hmmm, well it might be too low for this specific build/task execution. Try setting it much higher. "sudo sysctl fs.inotify.max_user_watches=524288" which is what my build machine has set. Dec 12 10:04:05 yep my machine has 524288 also Dec 12 10:05:04 tahnk you nrossi Dec 12 10:05:25 i will reboot now because i hate manually killing python processes Dec 12 10:05:42 i'll report back in about 30 minutes, hopefully Dec 12 10:05:42 odd my build machine at work has 8192? Dec 12 10:05:44 c0rnel: sysctl doesn't persist over reboot, so make sure to re-exec Dec 12 10:05:59 nrossi, i see Dec 12 10:06:19 what about systctl -n -w ? Dec 12 10:06:25 deepwater: Yer not sure tbh, could just be and spike due to specific ordering of files/tasks Dec 12 10:07:02 see you later Dec 12 10:07:04 bye Dec 12 10:07:05 it's worth a try I guess Dec 12 10:27:17 amazing Dec 12 10:27:27 the build seems to have actually started, now Dec 12 10:27:46 maybe it's somehow related to the older linux used (3.10) Dec 12 10:28:04 maybe scheduler is different, or whatever Dec 12 10:34:45 nrossi, thank you very much Dec 12 10:34:54 for now things seem to work Dec 12 10:35:17 hi Dec 12 10:35:27 anyone working withe meta-freescale? Dec 12 10:37:41 eikichi: probably - but usually its way more promising to state your *actual* question and the wait for rescue :) Dec 12 10:38:17 I don't have question about this Dec 12 10:38:43 eikichi: then the answer is "yes" Dec 12 10:38:47 except that there is no LICENCE so it doens't compile XD Dec 12 10:39:03 to be serious Dec 12 10:39:17 the meta-freescale repository has moved a lot recently Dec 12 10:39:34 meta-freescale is a new repository, I used to use meta-fsl-arm Dec 12 10:40:30 but I simply cannot move forward to it as Yocto checks for LICENCE that doesn't exist Dec 12 10:41:27 eikichi: file an issue, then? Dec 12 10:42:11 what are those LICENCES for= Dec 12 10:42:12 ? Dec 12 10:42:29 this is pretty annoying Dec 12 10:42:53 eikichi: i guess you are referring to some specific recipe inside the layer? Dec 12 10:43:30 eikichi: and licence files are there, well, to define the license of the several parts of code in use. they might be annoying to some, but for other they are absolutely crucial. Dec 12 10:45:09 yes Dec 12 10:45:12 this is the packagegroup-fsl-gstreamer1.0-full Dec 12 10:45:30 eikichi: meta-freescale repository is for morty Dec 12 10:46:10 older meta-fsl-* repositories disapeared Dec 12 10:46:20 https://github.com/Freescale/meta-fsl-arm Dec 12 10:57:41 weel Dec 12 10:57:44 well Dec 12 10:58:58 thanks, I'll stay on it Dec 12 12:52:54 hello, does anyone have experience with embedded Android... how does it compare to Yocto/OE? Dec 12 12:54:27 for what I've heard and read, it seems more limited, but my management thinks it could leverage the abundance of Java developers this way Dec 12 12:55:26 sounds like a fair assessment Dec 12 12:55:42 you'll be running on the android userspace so you're entirely at google's whim Dec 12 12:55:56 Initial board bringu-up will probably be more difficult/time consuming. But yes, there are lots more "app" people than "embedded sw engineers" Dec 12 12:56:40 also, I find there is less choice in terms of small SoMs that run proper embedded Android out of the box Dec 12 12:57:24 Ok Dec 12 12:57:40 actually the Variscite Snapdraon one is the only one I am aware of Dec 12 12:57:53 http://www.variscite.com/products/system-on-module-som/cortex-a53-krait/dart-sd410-qualcomm-snapdragon-410 Dec 12 13:00:34 also, has anyone tried Android Brillo Dec 12 13:00:40 ? Dec 12 13:01:35 as far as I understand its headless Android that basically drops the VM... Dec 12 13:01:41 If you're going for SoMs, you should take a look at SMARC and/or Qseven (https://www.sget.org/) Dec 12 13:02:58 JoiF, thanks...was actually not aware of those Dec 12 13:03:34 I actually keep a list of SoM vendors and embedded linux consulting companies just in case :) Dec 12 13:03:40 Should help you avoid vendor lock-in Dec 12 13:04:34 speaking of SoMs, I would actually prefer to put in work into our own board bring-up and BSP, but time to market is a pain as always Dec 12 13:04:35 I believe they're trying to make a standardized interface, similar to what PC-104 used to be Dec 12 13:05:13 JoiF: there are many standard. Dec 12 13:05:16 +s Dec 12 13:05:28 Yes. Dec 12 13:05:45 Of course there are ;) https://xkcd.com/927/ Dec 12 13:05:49 the problem is that I need the really small ones... like the DART6UL from Variscite Dec 12 13:06:08 and few Vendors make stuff in such a form factor Dec 12 13:06:32 Sounds like you need to roll your own, then :) Dec 12 13:07:40 handheld devices have very little space... if they'd let me stuff a Raspberry Pi inside them, I probably would :) Dec 12 13:08:28 Sounds like there's some requirements management missing there :) Dec 12 13:09:21 rburton: I could resend the piglit-compress-tests branch for M1 ... I guess I'll add RDEPENDS on tar based on previous discussion on ML Dec 12 13:09:40 List you you requirements formally, try: look for SoM (if applicable), purchase SoM, release_to_production. except NoSoMFoundError: roll_your_own. ;) Dec 12 13:11:40 jku: oh yes, please Dec 12 13:11:48 The process where I work is much simpler actually: We always roll our own. ;) Dec 12 13:13:06 if you only need to roll your som, thats ok. Dec 12 13:13:16 rolling a soc, on the other hand.... Dec 12 13:15:23 LetoThe2nd, I hope I live to see the day of desktop semiconductor manufacturing Dec 12 13:16:37 eduardas_m: we will see the day of no more desktop computing after all, except for special/professional reasons Dec 12 13:18:07 LetoThe2nd, year of the Linux desktop... the year when only greybreads use PCs...lol Dec 12 13:18:44 sry, greybeards :) Dec 12 13:20:03 nothing bad with greybread using linux. Dec 12 14:13:44 Hello, am I correct in assuming the genericx86 build is not capable of finding a rootfs on many boards, and I need to add a BSP for the architecture? Dec 12 14:28:25 Strike5150: it is not the genericx86 *machine* thats "not working" in that context, its in most cases the image type Dec 12 14:28:45 Strike5150: a bootable cd image should be fine basically. Dec 12 14:28:56 abelloni: did you ever find a nice way to make an am335x sd card image? preferably without root perms? Dec 12 14:32:11 Strike5150: i'd recommend not using genericx86 but something like meta-intel, if its compatible with your HW Dec 12 14:36:13 rburton: thats the better solution of course, but at least the ISO should boot Dec 12 14:37:39 agreed. there's a readme that explains how messed up x86 booting is Dec 12 14:37:51 you might need to faff about and make a usb stick that pretends to be a floppy disk Dec 12 14:38:22 really, one did manage to document all the bootup mess in *ONE* readme? Dec 12 14:38:29 Dec 12 14:38:32 :) Dec 12 14:38:48 well, there's a readme that says if nothing works, how to make a bootable usb stick that works differently Dec 12 14:38:57 Maybe, if it's 64-bit boot from EFI :-P Dec 12 14:40:15 Yea I'm basically at the point where it just can't find any disks that contain a rootfs, looking at meta-intel now and some of the wic scripts which I don't understand their purpose yet lol Dec 12 14:41:03 Ultimate goal is an image which boots on a single board computer with a super io chip and intel core Dec 12 14:47:24 Strike5150: for everything i have, hddimg works to boot off usb Dec 12 14:47:44 that's lame for production of course as its ext4 in a fat Dec 12 14:49:36 Yea, I tried the hddimg and it booted into u-boot and the kernel is ok, but it can't find the cf card slot Dec 12 14:50:00 CF card sucks but its what I have to work with hehe Dec 12 14:56:04 rburton: Why is ext4 on fat crap for production, just curious in what direction your thinking is going here. Dec 12 14:57:11 Strike5150: see if you can find a block diagram for the hardware to see how the CF card slot is connected and make sure the kernel is built with the driver config options required to support the hardware. You can add a bbappend to linux-yocto which adds a .cfg file containing the missing config options to SRC_URI. Dec 12 14:58:09 georgem: Right forgot to check those options, thanks Dec 12 14:58:48 Strike5150: because fat has a file size limit and going through a ext4 image on a fat partition is nonsense Dec 12 14:59:00 Strike5150: hddimg just does that for the live booting magic Dec 12 14:59:42 rburton: ahh yes ok ofcourse makes sense Dec 12 15:01:35 jstashluk: yes, using wic Dec 12 15:01:45 it is working failry well Dec 12 15:38:44 abelloni: I've been messing around with wic. Do you happen to have an example .wks that gets the partitions right? Dec 12 16:26:20 jstashluk: sdimage-bootpart is working fine Dec 12 16:28:59 don't forget to set IMAGE_BOOT_FILES Dec 12 16:34:07 jstashluk: wic works fine these days, try the samples in meta-yocto-bsp layer Dec 12 16:35:10 rburton: I have put together a glibc 2.25 upgrade, 2.25 is still pre-release but its coming out on Feb 1 (planned). Can you throw some AB cycles at it ? Dec 12 16:35:39 khem: next few days are going to be frantic testing for hte m1 release, but we can throw it at the AB immediate after that Dec 12 16:36:15 rburton: ok any timelines ? Dec 12 16:36:21 like next week ? Dec 12 16:36:36 later this week, hopefully Dec 12 16:36:47 rburton: and we should merge a pre-release into master early enough Dec 12 16:36:53 so we can fix fallouts Dec 12 16:36:55 yeah Dec 12 16:37:05 I have sent a fontconfig fix Dec 12 16:37:11 but I am sure there will be more Dec 12 16:37:17 your fix broke libxft :) Dec 12 16:37:23 as it changes API Dec 12 16:37:30 oh oh Dec 12 16:38:08 so this is not an issue which can be fixed surgicallu Dec 12 16:38:51 fontconfig has to break ABI I beleive they have done it in past to nothing new for that package Dec 12 16:39:19 i'd be a lot happpier merging patches that were backports for that though Dec 12 16:42:58 there are no backports yet Dec 12 16:43:13 I have sent a patch for fontconfig but I know it wont be applied as such Dec 12 16:43:20 however they now know the problem Dec 12 16:44:11 2.25 has deprecated inclusion of sys/sysmacros.h by sys/types.h Dec 12 16:44:26 so this will result in some pain Dec 12 16:45:22 rburton: how is gst 1.10 looking now ? Dec 12 17:49:34 hi guys , i would like to add altenative source . How do i do that . I gave SRC_URI=" https://a/b/c.tar.gz https://e/b/c.tar.gz" . Dec 12 17:49:46 but it always fails since first URL is down Dec 12 17:49:50 that's what MIRRORS is for Dec 12 17:49:59 https://a/b/c.tar.gz Dec 12 17:50:02 see the yocto product documentation, search for MIRRORS Dec 12 17:50:16 i did it , i couldn;t make out Dec 12 17:50:24 how to use ? Dec 12 17:50:58 i saw your branch gdb-linaro url is no more valid Dec 12 17:51:06 -SRC_URI = "https://releases.linaro.org/${MMYY}/components/toolchain/gdb-linaro/gdb-${PV}-${RELEASE}.tar.bz2" Dec 12 17:52:35 https://www.yoctoproject.org/docs/2.2/mega-manual/mega-manual.html#var-PREMIRRORS has an example of the syntax — the syntax of PREMIRRORS (which are checked before the url in SRC_URI) is the same as that of MIRRORS (which are checked after the url in SRC_URI) Dec 12 17:53:39 MIRRORS_append = " https://a/b/ https://e/b/ \n" woudl likely do, as would " https://.*/.* https://e/b \n" Dec 12 17:53:56 it's a search/replace. the url components matched by the first are replaced by the second Dec 12 17:54:04 using regular expression patterns applied to each part of the url Dec 12 17:55:35 "now you have two problems" Dec 12 17:55:37 thanks kergoth , i will check . Dec 12 17:56:00 seebs: :) Dec 12 17:56:47 dumb: you can probably grep around the layers to find examples of recipes which use MIRRORS. there aren't many, but should be some. i've done so myself, usually to handle the case where upstream moves the current release into an 'old' subdirectory for example, so it stops working when a new release comes out Dec 12 17:57:32 I set up mirrors on some perl recipes in meta-openembedded/meta-perl because CPAN was very flaky for a time Dec 12 17:58:50 thanks . i add missing dot . that is how yocto knows both SRC_URI is for fetching same package. Dec 12 17:59:05 I wanted to this in append file without modifying mainline Dec 12 17:59:12 if you put multiple urls in SRC_URI,,it'll always fetch both Dec 12 17:59:15 not a mirror setup Dec 12 17:59:40 if you needed extra sources rather than *alternative* sources, that would work Dec 12 18:00:49 correct . I will try few things , if that doesn't work . I will come back , thanks for your valuable input. Dec 12 18:25:15 Hello all, quick question: why standard/base applies patches for ARM when building powerpc? Dec 12 18:25:18 [ERROR]: Application of .kernel-meta//patches//arch/arm/v7-A15/ARM-LPAE-Invalidate-the-TLB-for-module-addresses-dur.patch failed. Dec 12 18:37:13 I have created my custom BSP layer using yocto-BSP. It's interesting to me why there are ARM patches that bitbake tries to apply - and obviously fails - when building for PowerPC platform. Dec 12 18:37:25 Did I miss something too obvious to catch? :D Dec 12 20:30:14 Heads up, there's a handful of long-pending bug reports (capset, python 3, and pthread mutex locking on fork) for which people have submitted patches to pseudo. Dec 12 20:30:28 These are now all merged in master. I haven't rerun every test, or anything, but they're pretty straightforward patches. Dec 12 20:31:21 Oh, wait. I missed one. Dec 12 20:33:03 okay, now the man page fix is in too. :) Dec 12 20:48:04 do you know maybe something about this patch (ARM-LPAE-Invalidate-the-TLB-for-module-addresses-dur.patch) failing to apply? Dec 12 20:48:16 or should I file a bug report? Dec 12 21:01:12 bug report filed: https://bugzilla.yoctoproject.org/show_bug.cgi?id=10792 Dec 12 21:01:14 Bug 10792: major, Undecided, ---, saul.wold, NEW , ARM-LPAE-Invalidate-the-TLB-for-module-addresses-dur.patch fails to apply Dec 12 21:26:41 zgredeq, looking at that log, it has nothing to do with the ARM patch; it is trying to create you a custom branch by reapplying all 366 patches that are already on standard/base. Dec 12 21:27:08 whatever you have in your custom machine doesn't have an association with a kernel branch. Dec 12 21:27:39 look at what happens if say you chose beaglebone... Dec 12 21:27:54 meta-poky/conf/local.conf.sample:#MACHINE ?= "beaglebone" Dec 12 21:28:21 the above would then make use of the defaults for that machine here: Dec 12 21:28:43 eta-yocto-bsp/recipes-kernel/linux/linux-yocto_4.8.bbappend:KBRANCH_beaglebone = "standard/beaglebone" Dec 12 21:28:43 meta-yocto-bsp/recipes-kernel/linux/linux-yocto_4.8.bbappend:SRCREV_machine_beaglebone ?= "28beb2f59f00de16b2cbdda20e9076c29f55b75a" Dec 12 21:28:44 meta-yocto-bsp/recipes-kernel/linux/linux-yocto_4.8.bbappend:COMPATIBLE_MACHINE_beaglebone = "beaglebone" Dec 12 21:28:44 meta-yocto-bsp/recipes-kernel/linux/linux-yocto_4.8.bbappend:LINUX_VERSION_beaglebone = "4.8.10" Dec 12 21:33:25 Hi paulg, thanks for answering. So basicaly, when yocto-bsp creates new BSP for me, there is a file called mycustombsp-standard.scc in recipes-kernel/linux/files Dec 12 21:34:44 among the lines are these ones: Dec 12 21:35:12 include ktypes/standard/standard.scc nopatch \ define KTYPE standard/base \ branch mycustommachine Dec 12 21:36:27 but as well there is a file called mycustommachine-tiny.scc Dec 12 21:37:04 include ktypes/tiny/tiny.scc \ define KTYPE standard/base \ branch mycustommachine Dec 12 21:37:26 one directory up you will find .bbappend file Dec 12 21:38:09 now, compiling standard with nopatch left intact produced error for me in do_package_qa saying that there was no GNU_HASH in the ELF binary in linux kernel... Dec 12 21:38:43 removing the "nopatch" fails the build, in the same way as selecting -tiny.scc instead of -standard.scc in .bbappend file Dec 12 21:42:42 either way, am I missing anything obvious here? Dec 12 21:50:41 well, it isn't clear to me how you are creating this BSP ; since you've already opened a bug, you probably want to put the exact steps there and not here in irc. Dec 12 21:52:30 ok, will do Dec 12 22:20:56 Is there a way to have a list of packages to build for an opkg repository that aren't included in the dist image? Dec 12 22:46:08 lolsborn: yes - you could use EXTRA_IMAGEDEPENDS or a packagegroup recipe to do that Dec 13 00:32:25 hi all, is there a possibility of having just the siginfo files as a separate sstate_mirror_url? Dec 13 00:33:03 meaning, tgz packages in one mirror_url and corresponding siginfo file in a different mirror url? Dec 13 00:33:13 no Dec 13 00:33:17 dear god why Dec 13 00:35:21 yes i hear you...some issue we are facing while hosting siginfo files Dec 13 00:47:09 in theory you could match by filename with sstate_mirrors.. but the entire idea is just horrible :) Dec 13 00:52:25 @kergoth: yes the idea is a horrid, we are facing some issue with siginfo file hence i was thinking on these lines Dec 13 00:52:40 but it doesn't seem to work Dec 13 00:57:22 the thing to remember about MIRRORS, PREMIRRORS, and SSTATE_MIRRORS is it isn't one regex pattern Dec 13 00:57:33 first it splits the url into its components, then *each component* is a regex search/replace Dec 13 02:54:45 hi does anyone know what is the mime setup for tgz.siginfo file in sstate-cache? Dec 13 02:55:03 we are having some trouble hosting sstate-cache.. **** ENDING LOGGING AT Tue Dec 13 03:00:00 2016