Hello,
Leap 42.2 Ring0 rebuilt for ARM is here: https://build.opensuse.org/project/monitor/home:matwey:experimental:Leap:42....
The following changes have to be introduced:
build: 1) apply https://github.com/openSUSE/obs-build/commit/2caf24a52292b0e23725d1821d48c4b... to fix build on armv7 (this is common issue also for Factory)
gcc5: 1) drop _constraints, they are never satisfied at armv7
libunwind: 1) apply c90a2e02b3c1b03362a549a05261a4d0513d6026 from upstream 2) apply 0bed10aa217157b287e036fba1e48ad2c6a3356e from upstream (fixed package is probably already in SLE queue)
post-build-checks: 1) apply https://github.com/openSUSE/post-build-checks/commit/c6e35c3b68457630fa73798...
What should I do with this changes? Should I submit them directly to Leap:42.2?
Leap 42.2 Ring1 is being built here: https://build.opensuse.org/project/monitor/home:matwey:experimental:Leap:42....
You are welcome to reuse my work or participate into fixing `failed' packages in Ring1.
Hi,
Le 03/05/2016 11:37, Matwey V. Kornilov a écrit :
Hello,
Leap 42.2 Ring0 rebuilt for ARM is here: https://build.opensuse.org/project/monitor/home:matwey:experimental:Leap:42....
The following changes have to be introduced:
build:
to fix build on armv7 (this is common issue also for Factory)
gcc5:
- drop _constraints, they are never satisfied at armv7
libunwind:
- apply c90a2e02b3c1b03362a549a05261a4d0513d6026 from upstream
- apply 0bed10aa217157b287e036fba1e48ad2c6a3356e from upstream (fixed
package is probably already in SLE queue)
post-build-checks:
What should I do with this changes? Should I submit them directly to Leap:42.2?
Leap 42.2 Ring1 is being built here: https://build.opensuse.org/project/monitor/home:matwey:experimental:Leap:42....
You are welcome to reuse my work or participate into fixing `failed' packages in Ring1.
Good job! :)
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
Guillaume
Hi Matwey, Guillaume,
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
I've used that tree for bootstrapping Leap:42.2:Ports/armv7l, thanks for those efforts! (it is still building). There are a few compile failures, I'd appreciate some help in fixing those.
Greetings, Dirk
2016-05-06 23:39 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey, Guillaume,
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
I've used that tree for bootstrapping Leap:42.2:Ports/armv7l, thanks for those efforts!
Great!
(it is still building). There are a few compile failures, I'd appreciate some help in fixing those.
There are some packages fixed in home:matwey:experimental:Leap:42.2:Rings:0. I need to fill bugreports and submit them to Leap. They are build, libunwind, post-build-checks and probably gcc5 (but I am not sure) In Rings:1 I found mariadb and boost with too tight _constraints for armv7l. I think we can use <override></override> here.
Greetings, Dirk
Hi Dirk,
Something strange with argyllcms package. Could you please trigger rebuild? Now it is failed, but when I branched it, it is built successfully: https://build.opensuse.org/package/show/home:matwey:branches:openSUSE:Leap:4... This package blocks a lot of stuff (gtk3.0, gnome, libzypp, etc).
2016-05-06 23:39 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey, Guillaume,
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
I've used that tree for bootstrapping Leap:42.2:Ports/armv7l, thanks for those efforts! (it is still building). There are a few compile failures, I'd appreciate some help in fixing those.
Greetings, Dirk
"Matwey V. Kornilov" matwey.kornilov@gmail.com writes:
Something strange with argyllcms package.
When it was last tried to be build, udev didn't exit yet, only udev-mini.
Andreas.
Hi Matwey,
Something strange with argyllcms package. Could you please trigger rebuild?
Yep, fixed now. Thanks!
Greetings, Dirk
Hi,
Could you please, remove the following from project config? It seems, that the following is not required.
# for bootstrap Support: !rpmlint-Factory-strict !rpmlint-Factory
At the same time, `build` is failed without rpmlint-Factory:
[ 79s] [ 0s] initializing /abuild/build-root/.srcfiles.cache ... [ 79s] [ 0s] /usr/src/packages/BUILD/obs-build-20151105/createdirdeps --oldfile /abuild/build-root/.srcfiles.cache /.build.binaries/ [ 80s] [ 0s] expanding package dependencies... [ 80s] [ 1s] Warning: distribution not specified, assuming 'default' (see /usr/src/packages/BUILD/obs-build-20151105/configs). [ 80s] [ 1s] expansion error [ 80s] [ 1s] nothing provides rpmlint-Factory
2016-05-10 14:08 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey,
Something strange with argyllcms package. Could you please trigger rebuild?
Yep, fixed now. Thanks!
Greetings, Dirk
Hi Matwey,
Could you please, remove the following from project config? It seems, that the following is not required.
# for bootstrap Support: !rpmlint-Factory-strict !rpmlint-Factory
Correct, those lines marked with "for bootstrap" were only there temorary to resolve a few bootstrapping issues. since we're almost complete I've removed them now and once I validated that the build config is proper, will trigger a full rebuild as currently packages are not signed correctly.
Greetings, Dirk
There is also currently common (in some packages) issue with gcc as the following:
[ 5803s] copymem_neon.o [ 5803s] In file included from /usr/include/features.h:402:0, [ 5803s] from ../../dist/system_wrappers/features.h:3, [ 5803s] from /usr/include/stdint.h:25, [ 5803s] from /usr/lib/gcc/armv7hl-suse-linux-gnueabi/4.8/include/stdint.h:9, [ 5803s] from ../../dist/system_wrappers/stdint.h:3, [ 5803s] from /usr/lib/gcc/armv7hl-suse-linux-gnueabi/4.8/include/arm_neon.h:39, [ 5803s] from /home/abuild/rpmbuild/BUILD/thunderbird/mozilla/media/libvpx/vp8/common/arm/neon/bilinearpredict_neon.c:11: [ 5803s] /usr/include/gnu/stubs.h:7:29: fatal error: gnu/stubs-soft.h: No such file or directory [ 5803s] # include <gnu/stubs-soft.h> [ 5803s]
Any ideas? Something is misconfigured in gcc probably.
2016-05-16 14:38 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey,
Could you please, remove the following from project config? It seems, that the following is not required.
# for bootstrap Support: !rpmlint-Factory-strict !rpmlint-Factory
Correct, those lines marked with "for bootstrap" were only there temorary to resolve a few bootstrapping issues. since we're almost complete I've removed them now and once I validated that the build config is proper, will trigger a full rebuild as currently packages are not signed correctly.
Greetings, Dirk
"Matwey V. Kornilov" matwey.kornilov@gmail.com writes:
There is also currently common (in some packages) issue with gcc as the following:
[ 5803s] copymem_neon.o [ 5803s] In file included from /usr/include/features.h:402:0, [ 5803s] from ../../dist/system_wrappers/features.h:3, [ 5803s] from /usr/include/stdint.h:25, [ 5803s] from /usr/lib/gcc/armv7hl-suse-linux-gnueabi/4.8/include/stdint.h:9, [ 5803s] from ../../dist/system_wrappers/stdint.h:3, [ 5803s] from /usr/lib/gcc/armv7hl-suse-linux-gnueabi/4.8/include/arm_neon.h:39, [ 5803s] from /home/abuild/rpmbuild/BUILD/thunderbird/mozilla/media/libvpx/vp8/common/arm/neon/bilinearpredict_neon.c:11: [ 5803s] /usr/include/gnu/stubs.h:7:29: fatal error: gnu/stubs-soft.h: No such file or directory [ 5803s] # include <gnu/stubs-soft.h> [ 5803s]
Any ideas? Something is misconfigured in gcc probably.
The arm soft-float ABI is not supported. Remove any use of -mfloat-abi=soft.
Andreas.
Hi all,
Where did VMinstall: kernel-obs-build come from in prjconf for openSUSE:Leap:42.2:Ports?
%ifarch %arm armv7l armv7hl VMinstall: kernel-obs-build %endif
If it is really needed, then it should be bootstrapped somehow.
2016-05-06 23:39 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey, Guillaume,
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
I've used that tree for bootstrapping Leap:42.2:Ports/armv7l, thanks for those efforts! (it is still building). There are a few compile failures, I'd appreciate some help in fixing those.
Greetings, Dirk
Hi Dirk,
Could you please take(aggregate) kernel-obs-build for openSUSE:Leap:42.2:Ports armv7l from home:matwey:experimental:Leap:42.2:Kernel QA_ports in order to bootstrap it in openSUSE:Leap:42.2:Ports? It also would be great if were able to trigger rebuild of all failed packages which failed with 'getbinaries: missing packages: kernel-obs-build'. Now, kernel-obs-build package is build blocker.
2016-05-29 15:17 GMT+03:00 Matwey V. Kornilov matwey.kornilov@gmail.com:
Hi all,
Where did VMinstall: kernel-obs-build come from in prjconf for openSUSE:Leap:42.2:Ports?
%ifarch %arm armv7l armv7hl VMinstall: kernel-obs-build %endif
If it is really needed, then it should be bootstrapped somehow.
2016-05-06 23:39 GMT+03:00 Dirk Müller dirk@dmllr.de:
Hi Matwey, Guillaume,
Once ring 1 will be in a good shape, I think we should move to an official project such as Leap:42.2:Ports or even Leap:42.2 with armv7l.
I've used that tree for bootstrapping Leap:42.2:Ports/armv7l, thanks for those efforts! (it is still building). There are a few compile failures, I'd appreciate some help in fixing those.
Greetings, Dirk
-- With best regards, Matwey V. Kornilov http://blog.matwey.name xmpp://0x2207@jabber.ru
Hi Matwey,
Could you please take(aggregate) kernel-obs-build for openSUSE:Leap:42.2:Ports
[...]
Thanks, I'm working on it (just returned from vacation today).
Greetings, Dirk
-
Andreas Schwab -
Dirk Müller -
Guillaume Gardet -
Matwey V. Kornilov