On Thu, Oct 19, 2017 at 09:16:06AM +0200, Arnd Bergmann wrote:
On Wed, Oct 18, 2017 at 11:45 AM, Greg KH gregkh@google.com wrote:
On Wed, Oct 18, 2017 at 09:56:57AM +0100, Milosz Wasilewski wrote:
On 18 October 2017 at 09:50, Linaro QA qa-reports@linaro.org wrote:
Summary
kernel: 4.4.92 git repo: https://android.googlesource.com/kernel/hikey-linaro git branch: android-hikey-linaro-4.4 git commit: f26ce5018d98da248eb149f2d3067a92186989e1 git describe: v4.4.92-568342-gf26ce5018d98 Test details: https://qa-reports.linaro.org/lkft/android-hikey-linaro-4.4-oe/build/v4.4.92...
No regressions (compared to build v4.4.91-568330-g4fe81adcc500)
Please note that LTP sched test suite is missing due to the following error:
Wait, how can this be "no regressions" when a test crashed the system? Seems like that should be reported somehow, right?
I suspect that is something we should reconsider. For a long time, some of the hardware was so unreliable that an incomplete test run was almost always an indication of lab problems rather than a kernel bug. One of the main problems (USB disconnect) is solved now, so if we've had a week of testing without any hardware faults, we may want to scream about any crashes in the short summary, at the risk of having a few false-positives.
The work we are doing to be able to assign log snippets to specific tests will also enable parsing the kernel log and generating an actual "test failure" when there is a kernel crash, and that will be picked up as a regression in cases like this.