[yocto] [meta-ti] Migration from 1.7.1 to 1.8 - kernel-abiversion missing
Brian Hutchinson
b.hutchman at gmail.com
Tue May 19 09:42:15 PDT 2015
On Tue, May 19, 2015 at 12:31 PM, Bruce Ashfield
<bruce.ashfield at windriver.com> wrote:
> On 2015-05-19 07:39 AM, Bruce Ashfield wrote:
>>
>> On Fri, May 15, 2015 at 4:21 PM, Brian Hutchinson <b.hutchman at gmail.com>
>> wrote:
>>>
>>> On Fri, May 15, 2015 at 3:26 PM, Brian Hutchinson <b.hutchman at gmail.com>
>>> wrote:
>>>>
>>>> On Fri, May 15, 2015 at 9:55 AM, Brian Hutchinson <b.hutchman at gmail.com>
>>>> wrote:
>>>>>
>>>>> On Thu, May 14, 2015 at 6:16 PM, Brian Hutchinson
>>>>> <b.hutchman at gmail.com> wrote:
>>>>>>
>>>>>>
>>>>>> On May 14, 2015 6:08 PM, "Denys Dmytriyenko" <denis at denix.org> wrote:
>>>>>>>
>>>>>>>
>>>>>>> On Tue, May 12, 2015 at 11:35:20AM -0400, Bruce Ashfield wrote:
>>>>>>>>
>>>>>>>> On 2015-05-12 10:20 AM, Brian Hutchinson wrote:
>>>>>>>>>
>>>>>>>>> On Mon, May 11, 2015 at 3:06 PM, Bruce Ashfield
>>>>>>>>> <bruce.ashfield at windriver.com> wrote:
>>>>>>>>>>
>>>>>>>>>> On 2015-05-11 02:10 PM, Brian Hutchinson wrote:
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Thu, Apr 30, 2015 at 10:06 AM, Bruce Ashfield
>>>>>>>>>>> <bruce.ashfield at windriver.com> wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> It is plausible. But in theory, linux-dummy should still provide
>>>>>>>>>>>> what you need (but since it doesn't build anything, there is
>>>>>>>>>>>> no abi .. and no modules can be built against it) .. so the
>>>>>>>>>>>> error isn't graceful.
>>>>>>>>>>>>
>>>>>>>>>>>> Bruce
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> I can confirm this same problem is happening to me. I just
>>>>>>>>>>> updated
>>>>>>>>>>> one of my builds from 1.7 to 1.8 and am also getting my rootfs to
>>>>>>>>>>> fail
>>>>>>>>>>> due to no abi kernel version:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> We still have a race condition in the 1.8 branch for the
>>>>>>>>>> population
>>>>>>>>>> of the build-artifacts directory.
>>>>>>>>>>
>>>>>>>>>> If modules start building, they'll race against the population of
>>>>>>>>>> the
>>>>>>>>>> abiversion, and you may see that message.
>>>>>>>>>>
>>>>>>>>>> There's a proposed patch for master, but I don't think it is in
>>>>>>>>>> fido yet.
>>>>>>>>>>
>>>>>>>>>> Bruce
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Hi Bruce,
>>>>>>>>>
>>>>>>>>> I did some searches and looks like there are a number of 'race'
>>>>>>>>> condition fixes but it wasn't obvious which one I may need. Is it
>>>>>>>>> this one:
>>>>>>>>
>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=02d0a003d603266114512160b209876199241e98
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> That's the one that should make sure that the shared workdir
>>>>>>>> (Which has the abiversion) is in place before building any modules.
>>>>>>>>
>>>>>>>> I can't say that it is exactly your issue, but it is the change
>>>>>>>> I was thinking of.
>>>>>>>
>>>>>>>
>>>>>>> Brian,
>>>>>>>
>>>>>>> Were you able to try the above mentioned commit against am180x in
>>>>>>> meta-ti?
>>>>>>> Did
>>>>>>> it solve the missing abi kernel version? Thanks.
>>>>>>>
>>>>>>> --
>>>>>>> Denys
>>>>>>
>>>>>>
>>>>>> Hi Denys,
>>>>>>
>>>>>> No, I got caught up in something else ... I'll try it tomorrow and
>>>>>> report
>>>>>> back after I cherry pick that commit Bruce mentioned.
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>> Brian
>>>>>
>>>>>
>>>>> Update. Not sure if I did this right but this is what I did. I added
>>>>> master as a remote and cherry picked
>>>>> 02d0a003d603266114512160b209876199241e98. Next I just went for it and
>>>>> tried to bitbake my image again and got the same result as before.
>>>>> Next I did a bitbake cleanall on virtual/kernel and tried to make my
>>>>> image again and still got the same result.
>>>>>
>>>>> I'm going to leave this build as is and setup a new one using 1.8
>>>>> master and see if I get the same thing again. I'll leave this broken
>>>>> build alone for a while in case someone wants me to try something with
>>>>> it to fix it.
>>>>>
>>>>> Regards,
>>>>>
>>>>> Brian
>>>>
>>>>
>>>> Yet another update ... I did a fresh checkout of master and tried to
>>>> build and had the same kernelabiversion error:
>>>>
>>>> WARNING: omap3-sgx-modules-5.01.01.01 ONLY supports hardfp mode for
>>>> now#######################
>>>> | ETA:
>>>> 00:00:28
>>>> WARNING: omap3-sgx-modules-5.01.01.02 ONLY supports hardfp mode for now
>>>> WARNING: ti-cgt6x-8.0.0 ONLY supports hardfp mode for
>>>> now########################################
>>>>
>>>> | ETA: 00:00:26
>>>> Parsing recipes: 100%
>>>>
>>>> |##############################################################################################################################################################################|
>>>> Time: 00:01:02
>>>> Parsing of 1802 .bb files complete (0 cached, 1802 parsed). 2303
>>>> targets, 182 skipped, 0 masked, 0 errors.
>>>> NOTE: Resolving any missing task queue dependencies
>>>> NOTE: multiple providers are available for u-boot (u-boot,
>>>> u-boot-glsdk, u-boot-ti-staging)
>>>> NOTE: consider defining a PREFERRED_PROVIDER entry to match u-boot
>>>> NOTE: multiple providers are available for jpeg (jpeg, libjpeg-turbo)
>>>> NOTE: consider defining a PREFERRED_PROVIDER entry to match jpeg
>>>>
>>>> Build Configuration:
>>>> BB_VERSION = "1.27.0"
>>>> BUILD_SYS = "x86_64-linux"
>>>> NATIVELSBSTRING = "Debian-7.8"
>>>> TARGET_SYS = "arm-poky-linux-gnueabi"
>>>> MACHINE = "am180x-evm"
>>>> DISTRO = "poky"
>>>> DISTRO_VERSION = "1.8+snapshot-20150515"
>>>> TUNE_FEATURES = "arm armv5 thumb dsp"
>>>> TARGET_FPU = "soft"
>>>> meta
>>>> meta-yocto
>>>> meta-yocto-bsp = "master:fab7da4f8030a4067db0522f77eaa6d3b501c68f"
>>>> meta-ti = "master:60a7bfbf96609ef6f3e084c32b2af853222b3b7e"
>>>> meta-oe
>>>> meta-python
>>>> meta-networking
>>>> meta-webserver = "master:53d55216c8c721d3b66ec8f968737bf081def870"
>>>>
>>>> NOTE: Preparing RunQueue
>>>> NOTE: Executing SetScene Tasks
>>>> NOTE: Executing RunQueue Tasks
>>>> WARNING: QA Issue: /usr/bin/apxs_apache2-dev contained in package
>>>> apache2-dev requires /usr/bin/perl, but no providers found in its
>>>> RDEPENDS [file-rdeps]
>>>> ERROR: No kernel-abiversion file found
>>>>
>>>> (/home/hutch/yocto_1.8_davinci_2/poky/build/tmp/sysroots/am180x-evm/pkgdata/kernel-depmod/kernel-abiversion),
>>>> cannot run depmod, aborting
>>>> ERROR: Function failed: do_rootfs
>>>> ERROR: Logfile of failure stored in:
>>>>
>>>> /home/hutch/yocto_1.8_davinci_2/poky/build/tmp/work/am180x_evm-poky-linux-gnueabi/core-image-nodeam/1.0-r0/temp/log.do_rootfs.10336
>>>> ERROR: Task 7
>>>> (/home/hutch/yocto_1.8_davinci_2/poky/meta/recipes-core/images/core-image-nodeam.bb,
>>>> do_rootfs) failed with exit code '1'
>>>> NOTE: Tasks Summary: Attempted 2614 tasks of which 9 didn't need to be
>>>> rerun and 1 failed.
>>>> Waiting for 0 running tasks to finish:
>>>>
>>>> Summary: 1 task failed:
>>>>
>>>> /home/hutch/yocto_1.8_davinci_2/poky/meta/recipes-core/images/core-image-nodeam.bb,
>>>> do_rootfs
>>>> Summary: There were 4 WARNING messages shown.
>>>> Summary: There were 2 ERROR messages shown, returning a non-zero exit
>>>> code.
>>>
>>>
>>> More info for those that care ...
>>>
>>> The end of the error log file has:
>>> DEBUG: Executing python function write_image_manifest
>>> DEBUG: Python function write_image_manifest finished
>>> NOTE: Executing: ldconfig
>>>
>>> -r/home/hutch/yocto_1.8_davinci_2/poky/build/tmp/work/am180x_evm-poky-linux-gnueabi/core-image-nodeam/1.0-r0/rootfs-c
>>> new -v
>>> ERROR: No kernel-abiversion file found
>>>
>>> (/home/hutch/yocto_1.8_davinci_2/poky/build/tmp/sysroots/am180x-evm/pkgdata/kernel-depmod/kernel-abiversion),
>>> cannot run depmod, aborting
>>> DEBUG: Python function do_rootfs finished
>>> ERROR: Function failed: do_rootfs
>>>
>>> I have a linux-dummy director in pkgdata but no kernel-depmod directory
>>> exists.
>>>
>>
>> Interesting. Looks like we have some sort of bad dependency. I'm poking
>> around
>> to see if I can reproduce this locally.
>
>
> Brian,
>
> I'm trying to wrap my head around this. Which kernel did you say
> was building in your configuration ? Was it really linux-dummy, or
> is there a bootable kernel in play ?
>
> Looking at the dependencies, I can see how the depmodwrapper requires
> that a kernel be built and the abiversion file created. But in the
> case of linux-dummy, there's no header files to use and generate the
> abiversion, so it won't be available for depmod.
>
> Which takes me back to one of my original questions, why is depmod
> running if the kernel isn't being built as part of this image .. and
> that leads me to think I'm missing some important information.
>
> Bruce
Hi Bruce,
Thanks for looking into it.
I have the meta-ti layer included and in the past, it did build a real
kernel version. I don't know when (because I don't use the kernel
built by the system, I use one from outside bitbake) but this release
(1.8) and last (1.7) has built linux-dummy instead of a real version
and I didn't tell it to.
Maybe I'm doing something wrong on my end but I built this release the
same way I've built all of them (and I've been doing it from the start
of yocto project).
The only other layers I use are meta-openembedded and meta-ti.
Regards,
Brian
More information about the yocto
mailing list