Summary ------------------------------------------------------------------------
kernel-repo: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git kernel-branch: linux-4.4.y kernel-commit: 573b59e17e37fc19577630391c08f7b1e26609d8 kernel-describe: v4.4.87 Test details: https://qa-reports.linaro.org/lkft/linux-stable-4.4-oe/build/v4.4.87
Regressions (compared to build v4.4.86) ------------------------------------------------------------------------
x86: kselftest: * test_align * test_progs
* Source: https://www.kernel.org/pub/linux/kernel/v4.x/linux-4.13.tar.xz
Documentation - https://collaborate.linaro.org/display/LKFT/Email+Reports
On 7 September 2017 at 20:24, Linaro QA qa-reports@linaro.org wrote:
Summary
kernel-repo: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git kernel-branch: linux-4.4.y kernel-commit: 573b59e17e37fc19577630391c08f7b1e26609d8 kernel-describe: v4.4.87 Test details: https://qa-reports.linaro.org/lkft/linux-stable-4.4-oe/build/v4.4.87
Regressions (compared to build v4.4.86)
x86: kselftest: * test_align * test_progs
* Source: https://www.kernel.org/pub/linux/kernel/v4.x/linux-4.13.tar.xz
It isn't a kernel regression. It's due to the kselftests changes. We moved from kselftests 4.12 + Shuah's next patches to ksefltests 4.13. I'm investigating the issue and will come back with more details.
Documentation - https://collaborate.linaro.org/display/LKFT/Email+Reports
-- Linaro QA (beta) https://qa-reports.linaro.org _______________________________________________ Lts-dev mailing list Lts-dev@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lts-dev
On Thu, Sep 07, 2017 at 05:24:39PM +0000, Linaro QA wrote:
Summary
kernel-repo: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git kernel-branch: linux-4.4.y kernel-commit: 573b59e17e37fc19577630391c08f7b1e26609d8 kernel-describe: v4.4.87 Test details: https://qa-reports.linaro.org/lkft/linux-stable-4.4-oe/build/v4.4.87
Regressions (compared to build v4.4.86)
x86: kselftest: * test_align * test_progs
* Source: https://www.kernel.org/pub/linux/kernel/v4.x/linux-4.13.tar.xz
Is that the source of kselftest? Or the kernel build?
And did the build itself succeed? What configurations are you building for x86?
And finally, why are you testing x86, what happened to the ARM platforms?
thanks,
greg k-h
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
On Thu, Sep 07, 2017 at 05:24:39PM +0000, Linaro QA wrote:
Summary
kernel-repo: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git kernel-branch: linux-4.4.y kernel-commit: 573b59e17e37fc19577630391c08f7b1e26609d8 kernel-describe: v4.4.87 Test details: https://qa-reports.linaro.org/lkft/linux-stable-4.4-oe/build/v4.4.87
Regressions (compared to build v4.4.86)
x86: kselftest: * test_align * test_progs
* Source: https://www.kernel.org/pub/linux/kernel/v4.x/linux-4.13.tar.xzIs that the source of kselftest? Or the kernel build?
Source of kselftest. It has already been renamed to "* test src:" in the email template.
And did the build itself succeed?
The build of? kselftests? There's a few failures. That's what I'm currently investigating.
What configurations are you building for x86?
x86_64_defconfig + config fragments from in-kernel selftests. The result is here: http://snapshots.linaro.org/openembedded/lkft/morty/intel-core2-32/rpb/linux...
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
thanks,
greg k-h _______________________________________________ Lts-dev mailing list Lts-dev@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lts-dev
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
But having said that, I clicked on Test details and didn't see hikey tested. There's one kernel config listed in the metadata which is x86-specific, so I'm confused...
Hi Steve, Greg,
On Sep 8, 2017 02:41, "Steve Muckle" smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
But having said that, I clicked on Test details and didn't see hikey tested. There's one kernel config listed in the metadata which is x86-specific, so I'm confused...
Sorry, 4.4.y is a 'special' case wrt Hikey: since Hikey support wasn't merged until later, to test LTS-4.4, we have a hikey-merged-4.4.h which results into another test run called linaro-hikey-stable-4.4-oe (likewise for -rc).
That's why, this particular test build only tests x86 at the moment.
I know it is a bit confusing, but keeping a 'vanilla' stable-4.4 setup allows us a comparison benchmark for each run.
Hope this helps.
Best, Sumit.
_______________________________________________ Lts-dev mailing list Lts-dev@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lts-dev
On Fri, Sep 08, 2017 at 08:34:48AM +0530, Sumit Semwal wrote:
Hi Steve, Greg,
On Sep 8, 2017 02:41, "Steve Muckle" smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote: On 7 September 2017 at 22:10, Greg KH <gregkh@google.com> wrote: And finally, why are you testing x86, what happened to the ARM platforms? Beagle X15 is being added to the lab. The results should be available soon. I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)? This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed. But having said that, I clicked on Test details and didn't see hikey tested. There's one kernel config listed in the metadata which is x86-specific, so I'm confused...Sorry, 4.4.y is a 'special' case wrt Hikey: since Hikey support wasn't merged until later, to test LTS-4.4, we have a hikey-merged-4.4.h which results into another test run called linaro-hikey-stable-4.4-oe (likewise for -rc).
That's why, this particular test build only tests x86 at the moment.
I know it is a bit confusing, but keeping a 'vanilla' stable-4.4 setup allows us a comparison benchmark for each run.
Will we ever get a "vanilla" setup for 4.4 kernels on a non-x86 system? And given the past failures of tests on arm64 only, it's not that good of a comparison benchmark :)
How about for 4.9? 4.14?
thanks,
greg k-h
On Fri, Sep 08, 2017 at 11:36:26AM +0200, Greg KH wrote:
Will we ever get a "vanilla" setup for 4.4 kernels on a non-x86 system? And given the past failures of tests on arm64 only, it's not that good of a comparison benchmark :)
There's plenty of ARM boards that work well upstream for older kernels (even way earlier than v4.4) but unfortunately there was an insistance from the Android point of view on using HiKey as the lead board since it's a public board that's in AOSP.
I understand that the x15 that's currently being added to LKFT is supposed to be fine on v4.4 and will give 32 bit coverage, I've also been asking for Juno which would give arm64 coverage there (it's what the kernel team at Linaro mostly use internally for LTS stuff), it's on the list for the team but I'm not sure when they're planning to add it. The Dragonboard 410c will run vanilla v4.4 as well, I think that's on the list too.
How about for 4.9? 4.14?
HiKey finally got things mostly working upstream in v4.9 so that should be fine, John Stultz and the rest of the Android team at Linaro put a lot of effort into that.
On 8 September 2017 at 13:07, Mark Brown broonie@kernel.org wrote:
On Fri, Sep 08, 2017 at 11:36:26AM +0200, Greg KH wrote:
Will we ever get a "vanilla" setup for 4.4 kernels on a non-x86 system? And given the past failures of tests on arm64 only, it's not that good of a comparison benchmark :)
There's plenty of ARM boards that work well upstream for older kernels (even way earlier than v4.4) but unfortunately there was an insistance from the Android point of view on using HiKey as the lead board since it's a public board that's in AOSP.
I understand that the x15 that's currently being added to LKFT is supposed to be fine on v4.4 and will give 32 bit coverage, I've also
x15 boards are already available in the lab. Dave runs stress testing to check for possible infrastructure problems. I expect them to be available for regular testing next week.
been asking for Juno which would give arm64 coverage there (it's what the kernel team at Linaro mostly use internally for LTS stuff), it's on the list for the team but I'm not sure when they're planning to add it. The Dragonboard 410c will run vanilla v4.4 as well, I think that's on the list too.
the boards are in the lab running stress tests as well. So there is a chance they could be available soon. The problem is I don't see a build for it. IIRC they didn't work on 4.4 but I might have missed something.
How about for 4.9? 4.14?
HiKey finally got things mostly working upstream in v4.9 so that should be fine, John Stultz and the rest of the Android team at Linaro put a lot of effort into that.
hikeys work on 4.9. Here are the last results I have: https://qa-reports.linaro.org/lkft/linux-stable-4.9-oe/build/v4.9.47/
I'm now re-testing 4.9.48 as there was a small problem with test jobs (after last template changes)
milosz
On Fri, Sep 08, 2017 at 01:19:31PM +0100, Milosz Wasilewski wrote:
On 8 September 2017 at 13:07, Mark Brown broonie@kernel.org wrote:
The Dragonboard 410c will run vanilla v4.4 as well, I think that's on the list too.
the boards are in the lab running stress tests as well. So there is a chance they could be available soon. The problem is I don't see a build for it. IIRC they didn't work on 4.4 but I might have missed something.
They've got support upstream which wasn't really feature complete but I thought was good enough for a lot of the things we're doing with LTS, the DTS certainly looks promising. ICBW though, it won't be at all useful for Android but that needs an awful lot more hardware support.
On Fri, Sep 08, 2017 at 01:07:53PM +0100, Mark Brown wrote:
On Fri, Sep 08, 2017 at 11:36:26AM +0200, Greg KH wrote:
Will we ever get a "vanilla" setup for 4.4 kernels on a non-x86 system? And given the past failures of tests on arm64 only, it's not that good of a comparison benchmark :)
There's plenty of ARM boards that work well upstream for older kernels (even way earlier than v4.4) but unfortunately there was an insistance from the Android point of view on using HiKey as the lead board since it's a public board that's in AOSP.
I should say /the/ public board rather than /a/ public board - it's really the only currently viable option for Android testing, though it's a problem from an upstream point of view in v4.4.
On 7 September 2017 at 22:10, Steve Muckle smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
It's not for architecture (yet) but for device. Right now we happen to have 1 device for each architecture. I have grouping by arch on my todo list.
milosz
But having said that, I clicked on Test details and didn't see hikey tested. There's one kernel config listed in the metadata which is x86-specific, so I'm confused... _______________________________________________ Kernel-build-reports mailing list Kernel-build-reports@lists.linaro.org https://lists.linaro.org/mailman/listinfo/kernel-build-reports
On Fri, Sep 08, 2017 at 10:16:00AM +0100, Milosz Wasilewski wrote:
On 7 September 2017 at 22:10, Steve Muckle smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
It's not for architecture (yet) but for device. Right now we happen to have 1 device for each architecture. I have grouping by arch on my todo list.
So does that mean we will be getting one email per device, and not just one email per release?
If so, please reconsider...
greg k-h
On Fri, Sep 08, 2017 at 11:31:55AM +0200, Greg KH wrote:
On Fri, Sep 08, 2017 at 10:16:00AM +0100, Milosz Wasilewski wrote:
On 7 September 2017 at 22:10, Steve Muckle smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
It's not for architecture (yet) but for device. Right now we happen to have 1 device for each architecture. I have grouping by arch on my todo list.
So does that mean we will be getting one email per device, and not just one email per release?
If so, please reconsider...
And finally, why are these all being cc:ed to a mailing list that bounces whenever a non-linaro member posts to them (i.e. qa-reports@linaro.org)?
I like lots of email and all that, but delivery failures for every one I send out is not nice.
thanks,
greg k-h
On 8 September 2017 at 10:37, Greg KH gregkh@google.com wrote:
On Fri, Sep 08, 2017 at 11:31:55AM +0200, Greg KH wrote:
On Fri, Sep 08, 2017 at 10:16:00AM +0100, Milosz Wasilewski wrote:
On 7 September 2017 at 22:10, Steve Muckle smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
It's not for architecture (yet) but for device. Right now we happen to have 1 device for each architecture. I have grouping by arch on my todo list.
So does that mean we will be getting one email per device, and not just one email per release?
If so, please reconsider...
And finally, why are these all being cc:ed to a mailing list that bounces whenever a non-linaro member posts to them (i.e. qa-reports@linaro.org)?
I like lots of email and all that, but delivery failures for every one I send out is not nice.
I know, I'm trying to fix it now.
milosz
On 8 September 2017 at 10:31, Greg KH gregkh@google.com wrote:
On Fri, Sep 08, 2017 at 10:16:00AM +0100, Milosz Wasilewski wrote:
On 7 September 2017 at 22:10, Steve Muckle smuckle.linux@gmail.com wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
It's not for architecture (yet) but for device. Right now we happen to have 1 device for each architecture. I have grouping by arch on my todo list.
So does that mean we will be getting one email per device, and not just one email per release?
If so, please reconsider...
1 email per commit/release of each tree. As Sumit noted hikey is not booting with vanilla 4.4 so it's build from a different tree.
What I meant is that the regressions are grouped by devices and we happen to have 1 device for each architecture. So it might look like we're grouping by architecture.
milosz
On Thu, Sep 07, 2017 at 02:10:05PM -0700, Steve Muckle wrote:
On 09/07/2017 12:28 PM, Fathi Boudra wrote:
On 7 September 2017 at 22:10, Greg KH gregkh@google.com wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
Beagle X15 is being added to the lab. The results should be available soon.
I think Greg may be referring to ARM[32/64] in general (Greg please correct me if not)?
This is now just listing regressions from past test results, so I'm assuming that if there are no regressions for a particular architecture, that architecture will not be listed.
If so, that's not going to be very helpful as ideally I want to know that I didn't break anything when I do a -rc release, so I know it is safe to release it.
And as I have said before, x86 isn't interesting, as that is easy, and others are already doing that work...
thanks,
greg k-h
On Thu, Sep 07, 2017 at 09:10:27PM +0200, Greg KH wrote:
And finally, why are you testing x86, what happened to the ARM platforms?
The main reason x86 is covered is that it's helpful to be able to do the comparison to see if things are architecture specific and also that it's helpful when reporting things upstream to be able to show something that runs on a more widely available architecture.
-
Fathi Boudra -
Greg KH -
Linaro QA -
Mark Brown -
Milosz Wasilewski -
Steve Muckle -
Sumit Semwal