Hello John and Martin,

On Tue, Jan 26, 2021 at 5:31 PM Martin Jansa <[email protected]> wrote:
>
> Yes, it's related to rebuilding in the same WORKDIR and the only fix which is 
> needed to be backported to dunfell should be:
> 04afc692a gpsd: mark CLEANBROKEN
>
> see the previous discussion about this issue:
> https://lists.openembedded.org/g/openembedded-devel/topic/74225704#84379

Actually, I remember I was reading this and put it on my "internal
watch list", but quickly dropped it since the solution was provided!

Martin,
Would it be possible to cherry-pick that commit onto dunfell to fix
those build errors then?

>
> On Tue, Jan 26, 2021 at 7:58 AM John Ernberg <[email protected]> wrote:
>>
>> Hi Andrey, Steve,
>>
>> On 1/26/21 4:25 PM, Andrey Zhizhikin wrote:
>> > Hello Steve,
>> >
>> > On Tue, Jan 26, 2021 at 3:29 PM Steve Sakoman <[email protected]> wrote:
>> >>
>> >> Hi Andrey,
>> >>
>> >> Thanks for reporting this issue.
>> >>
>> >> I'm the maintainer for the dunfell branch of oe-core.  The gpsd recipe
>> >> is part of meta-openembedded. I'm copying Armin, who is the maintainer
>> >> for the dunfell branch of meta-openembedded.
>> >
>> > True, I've realized the recipe is not part of OE-Core.
>> >
>> >>
>> >> That said, I'm surprised you are seeing this issue since the gpsd
>> >> recipe has been fairly stable for quite some time.  I know that Armin
>> >> does test builds quite frequently.
>> >>
>> >> FWIW, I tried doing a gpsd build with oe-core, meta-openembedded,
>> >> meta-freescale, and meta-freescale-distro on an Ubuntu 20.04 and it
>> >> completed successfully.
>> >
>> > Arghh, got the same setup now - and the build issue was gone! :( Looks
>> > like this was an intermittent one
>> >
>> > I went ahead and invalidated the sstate for the recipe, and build went
>> > through alright.
>> >
>> > With that said: please discard this message, and thanks for responding 
>> > here!
>> >
>> >>
>> >> Perhaps more information on your environment might help.  What
>> >> distro/version is the OS on your build machine? What other layers are
>> >> enabled in your build? What are you using for MACHINE and DISTRO?  Do
>> >> any of your layers have a .bbappend for gpsd?
>> >
>> > For the record: had nothing fancy, same setup as you've indicated above:
>> >
>> > Build Configuration:
>> > BB_VERSION           = "1.46.0"
>> > BUILD_SYS            = "x86_64-linux"
>> > NATIVELSBSTRING      = "universal"
>> > TARGET_SYS           = "aarch64-fsl-linux"
>> > MACHINE              = "imx8mmevk"
>> > DISTRO               = "fsl-wayland"
>> > DISTRO_VERSION       = "3.1"
>> > TUNE_FEATURES        = "aarch64 cortexa53 crc crypto"
>> > TARGET_FPU           = ""
>> > meta
>> > meta-poky            = "HEAD:bb7747497adbc7c99f6fc9b48b643eecb4cb1408"
>> > meta-oe
>> > meta-multimedia
>> > meta-python
>> > meta-networking
>> > meta-filesystems     = "dunfell:5bba79488b7d393d2258d6e917f7bf7b0d7c4073"
>> > meta-freescale       =
>> > "update-kernel-dunfell:184ee6b12d1c06db9327535ad1ae84d4709cfdb0"
>> > meta-freescale-3rdparty = "HEAD:ed841161a97307ebd901c31c62f8ecbee6baaacf"
>> > meta-freescale-distro = "HEAD:5d882cdf079b3bde0bd9869ce3ca3db411acbf3b"
>> >
>> >>
>> >> Regards,
>> >>
>> >> Steve
>> >>
>> >> On Tue, Jan 26, 2021 at 2:13 AM Andrey Zhizhikin <[email protected]> 
>> >> wrote:
>> >>>
>> >>> Hello Steve,
>> >>>
>> >>> gpsd recipe build breaks on the dunfell branch with following error 
>> >>> messages:
>> >>>
>> >>> | 
>> >>> /development/dunfell/build-output/work/aarch64-fsl-linux/gpsd/3.19-r0/recipe-sysroot-native/usr/bin/aarch64-fsl-linux/../../libexec/aarch64-fsl-linux/gcc/aarch64-fsl-linux/9.3.0/ld:
>> >>> ./libgpsd.a(ppsthread.o): undefined reference to symbol
>> >>> 'pthread_create@@GLIBC_2.17'
>> >>> | 
>> >>> /development/dunfell/build-output/work/aarch64-fsl-linux/gpsd/3.19-r0/recipe-sysroot-native/usr/bin/aarch64-fsl-linux/../../libexec/aarch64-fsl-linux/gcc/aarch64-fsl-linux/9.3.0/ld:
>> >>> /development/dunfell/build-output/work/aarch64-fsl-linux/gpsd/3.19-r0/recipe-sysroot/lib/libpthread.so.0:
>> >>> error adding symbols: DSO missing from command line
>>
>> We have seen this one as well, we have not identified why it happens,
>> but we believe it is triggered by an incremental rebuild of the recipe.

Thanks for a quick confirmation here, it's good to know I was not
alone to observe this issue. :)

>>
>> We have enabled systemd for gpsd, and if systemd is rebuilt due to a
>> patch or whatever, gpsd appears to error out exactly like this.
>>
>> A very quick analysis we did when it first happened seem to indicate
>> that Scons (when using a cache) gets confused about the pthread
>> capability in the environment.
>>
>> We have not been able to research the error further yet.
>>
>> Best regards // John Ernberg
>> >>>
>> >>> This cannot be reproduced on the current master branch, where the
>> >>> version of gpsd recipe has been bumped up from 3.19 to 3.20.
>> >>>
>> >>> Can you please consider to cherry-pick following commits from master
>> >>> in order to upgrade the gspd version on dunfell branch:
>> >>>
>> >>> 98e61961c gpsd: add new init script
>> >>> 04afc692a gpsd: mark CLEANBROKEN
>> >>> 33a15c790 gpsd: fix build on aarch64
>> >>> cecab9c4c gpsd: Upgrade 3.19 -> 3.20
>> >>>
>> >>> I've tried to cherry-pick them locally and they do apply clean. Also
>> >>> the build of gpsd is successful after this upgrade.
>> >>>
>> >>> --
>> >>> Regards,
>> >>> Andrey.
>> >
>> >
>> > --
>> > Regards,
>> > Andrey.
>> >
>> >
>> >
>> >
>> >
>>
>>
>>
>
> 
>


-- 
Regards,
Andrey.
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#147317): 
https://lists.openembedded.org/g/openembedded-core/message/147317
Mute This Topic: https://lists.openembedded.org/mt/80128605/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to